Professional Documents
Culture Documents
Emerging Technologies For Education
Emerging Technologies For Education
Emerging Technologies For Education
Emerging Technologies
for Education
4th International Symposium, SETE 2019
Held in Conjunction with ICWL 2019
Magdeburg, Germany, September 23–25, 2019
Revised Selected Papers
Lecture Notes in Computer Science 11984
Founding Editors
Gerhard Goos
Karlsruhe Institute of Technology, Karlsruhe, Germany
Juris Hartmanis
Cornell University, Ithaca, NY, USA
Emerging Technologies
for Education
4th International Symposium, SETE 2019
Held in Conjunction with ICWL 2019
Magdeburg, Germany, September 23–25, 2019
Revised Selected Papers
123
Editors
Elvira Popescu Tianyong Hao
University of Craiova South China Normal University
Craiova, Romania Guangzhou, China
Ting-Chia Hsu Haoran Xie
National Taiwan Normal University Lingnan University
Taipei, Taiwan Hong Kong, Hong Kong
Marco Temperini Wei Chen
Sapienza University of Rome Chinese Academy of Agricultural Sciences
Rome, Italy Beijing, China
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
Prof. Michael Herzog (especially Veronika Weiß and Leonore Franz), for their efforts
and time spent to ensure the success of the conference.
General Co-chairs
Elvira Popescu University of Craiova, Romania
Tianyong Hao South China Normal University, China
Workshop Co-chairs
Marco Temperini Sapienza University, Italy
Wei Chen Chinese Academy of Agricultural Sciences, China
Publicity Chair
Liang Chen Fujian Normal University, China
Track Co-chairs
T01: Emerging Technologies for Open Access to Education and Learning
Gabriela Grosseck West University of Timisoara, Romania
Gwo-Jen Hwang National Taiwan University of Science
and Technology, Taiwan
Workshop Organizers
The 12th International Workshop on Social and Personal Computing
for Web-Supported Learning Communities (SPeL 2019)
Elvira Popescu University of Craiova, Romania
Sabine Graf Athabasca University, Canada
Organization ix
A Case Study on How Greek Teachers Make Use of Big Data Analytics
in K-12 Education. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Anna Mavroudi and Spyros Papadakis
papadakis@eap.gr
Abstract. Big Data Analytics can help teachers to make better and informed
decisions. Several recent articles in the field of technology enhanced learning
concern this potential, yet little is known about how teachers actually make use
of Big Data Analytics in their school to support themselves and their students.
To compensate for this gap, this paper focuses on the actual uses of Big Data
Analytics by active schoolteachers. Thirty teachers who live in Greece participated
in survey about their usage of (a) Big Data analytics and (b) online learning
environments which capture student data. The data were analysed using mixed
methods. Main findings reveal that the schoolteachers are storing and actively
using student data as well as Big Data which involve the support of the teaching-
learning process. Also, it became clear that teachers use Big Data Analytics for
two main distinctively different purposes: to cover teaching-learning aspects and to
complete administrative tasks. Finally, it emerged that a small number of teachers is
archiving digital multimedia. Consequently, a need arises for appropriate analytics
and relevant privacy frameworks. Other practical implications of the findings of
this work touch upon the design of teachers ‘development programs in Big Data
and their analytics.
1 Introduction
The abundance of digital traces that students are leaving in the digital learning envi-
ronments coupled with advances in artificial intelligence, cloud computing, machine
learning and data mining, has generated new opportunities in the educational technol-
ogy field [16, 17], which are epitomised in the support of “evidence-based learning”.
The growing popularity of online or blended learning coupled with “evidence-based
learning” approaches has contributed to the rise of the field of Big Data (BD) Analytics
in Education [2]. It exploits a set of tools that can provide insights which can facilitate
pedagogical decision-making [18]. Yet, it has been argued that we need to better under-
stand the users’ perspectives [7], e.g. students and teachers. This paper focuses on the
current uses of BD by teachers in the context of Greek school education. School educa-
tion is a context which is heavily underrepresented in this topic at the recent literature
[7]. The aim of this study is to understand what types of BD schoolteachers use, where
are these data coming from, and how they use them.
2 Background
2.1 Big Data in Education
One common definition is that BD is data stored in “large and complex datasets col-
lected from digital and conventional sources that are not easily managed by traditional
applications or processes” [16, p. 75]. There is a strong link between BD in Education
and Learning Analytics summarised as follows: “the process of gathering, analyzing
and reporting educational big data is referred to as learning analytics (LA)” [16, p. 76].
A recent review on the application of BD in education reveals several different peda-
gogical cases of potential exploitation, including performance prediction, attrition risk
detection, data visualisation, intelligent feedback, course recommendation, student skills
estimation, behaviour detection, grouping and collaboration of students [17]. One of the
sources of BD in Education comes from the user tracking and monitoring capabilities
of online Learning Management Systems (LMS).
3 Method
3.1 Instrument
A questionnaire was used which was comprised of three main categories of questions:
basic participants’ demographics, participants’ usage of BD analytics (related either
A Case Study on How Greek Teachers Make Use of Big Data Analytics 5
Teachers who are active in primary or secondary education in Greece were contacted via
email asking them to complete the questionnaire anonymously. The link to the online
questionnaire was included in the email message. Another prerequisite regarding the
teachers’ participation in the research was that their school was storing and using dig-
ital archives that contain student data. After a time period of a month, the answers
were collected and analysed. The demographics data were analysed manually. The text
that corresponded to the participants’ answers in the open-ended questions was anal-
ysed using a software for qualitative analyses and content analysis as the main analysis
method.
4 Results
In total, 30 teachers answered the questionnaire, 13 females and 17 males. One third of
them works in primary education while the remaining two thirds are active in secondary
education. With respect to age, one participant belongs in the 21–35 age group, nine
participants belong to the 36–45 age group and the remaining twenty participants belong
to the 46–65 age group. All participants declared that their school is storing and actively
using digital archives containing student data; also, the majority of them (22 out of
30) declared that their school does the same for data which involve the support of the
teaching-learning process. In this question, four participants answered negatively and
the remaining four answered that they were not aware of whether such data are being
collected in their school unit.
With respect to the student data kept in a digital format, Table 1 depicts the main
data categories and the associated frequency of occurrence in the participants’ answers.
Grades and students’ attendance are the data types kept most frequently, along with
personal data like contact details (address and phone), and parents’ names. In addition, it
seems that the teachers capitalise on the affordances of educational technology systems in
keeping in a digital format other student data such as: progress, awards, artifacts (created
6 A. Mavroudi and S. Papadakis
Following, in the question “in what ways are these data used?” the main answer in
ten cases is quite generic (e.g. “in ways that help us covering the needs of the school
unit”); or variations of that answer, without specifying more though. These variations
revolve around: better organisation of the school life, educational, administrative or
functional needs of the school. Four cases point out that they are keeping student data in
line with national regulations (in one case this is explicitly mentioned) and consequently
they are used in accordance with it. For example, to monitor absence of students from
school, to issue grade certificates at the end of the semester or certificates of completion
of an educational program. Other reasons mentioned are: monitoring of and providing
feedback in the teaching-learning process, as well as archiving decisions made by the
local educational policymakers.
With respect to online educational technology tools or resources used, Fig. 1 depicts
the teachers’ answers. The “MySchool” environment and online OERs are the most
popular options among teachers, since they are supported by the Ministry of Education
and are freely available. Following, four types of learning environments received the
same attention from the participant teachers: the use of social media, LMS, Google
Docs or similar, and applications that can support mobile learning. Less popular is the
use of e-portfolios and finally, a considerate percentage of teachers makes use of other,
additional tools.
Finally, the question “how do you use online educational technology tools?” was ask-
ing for more in-depth information. The response rate in this question was low, receiving
13 answers in total. Some answers were describing tool-specific usage, whereas others
were more abstract and orientated towards the description of pedagogical methods that
A Case Study on How Greek Teachers Make Use of Big Data Analytics 7
can be supported by the educational technology tools. Two main emerging themes are
the improvement of the teaching-learning process and the administration of education.
Examples of the former include visual learning, practice knowledge via test items, and
communication skills development. Examples of the latter include communication with
parents, and improving the visibility of the school activities among stakeholders. These
examples were mentioned by the participants.
both for the research community and the policymakers. The contribution of this work
relates to the fact that there is lack of empirical evidence on the topic at stake in the
context of K-12 education in the literature, while the usage of a new technology by the
practitioners is a crucial parameter when designing a strategy for uptake at scale.
Also, the research implications of this work touch upon the recommendations for
future research, while the practical implications upon the usage of BD Analytics in
schools as a proxy for their readiness to effectively engage with this learning innovation.
In turn, this insight can be useful in the design of teachers’ professional development
programs. For instance, it could be argued that no evidence was found herein with
respect to using BD Analytics as a means of promoting high-order thinking skills among
students, e.g. as a metacognitive tool; something which is recommended in the recent
relevant literature [5]. Consequently, teachers’ training programs on BD Analytics could
try to address this gap. With respect to societal implications, the most frequently used
BD Analytics among the participant teachers was students’ grades and class attendance.
The literature suggests that these two indicators combined have strong correlation with
the identification of student at-risk of dropping out [10] which is in line with our findings.
Finally, the need for updated privacy frameworks for multimodal BD Analytics has also
emerged both herein and in the literature [14].
References
1. Advanced Electronic Scenarios Operating Platform (A.E.S.O.P.). http://aesop.iep.edu.gr.
Accessed 9 June 2019
2. Cope, B., Kalantzis, M.: Sources of evidence-of-learning: learning and assessment in the era
of big data. Open Rev. Educ. Res. 2(1), 194–217 (2015)
3. e-learning service of GSN homepage. http://e-learning.sch.gr. Accessed 9 June 2019
4. Electronic Classroom service homepage. http://eclass.sch.gr. Accessed 9 June 2019
5. Gašević, D., Dawson, S., Siemens, G.: Let’s not forget: learning analytics are about learning.
TechTrends 59(1), 64–71 (2015)
6. Greek School Network homepage. https://www.sch.gr/. Accessed 9 June 2019
7. Li, K.C., Lam, H.K., Lam, S.S.: A review of learning analytics in educational research. In:
Lam, J., Ng, K.K., Cheung, S.K.S., Wong, T.L., Li, K.C., Wang, F.L. (eds.) ICTE 2015.
CCIS, vol. 559, pp. 173–184. Springer, Heidelberg (2015). https://doi.org/10.1007/978-3-
662-48978-9_17
8. Learning Activity Management System (LAMS) foundation homepage. https://www.
lamsfoundation.org/. Accessed 9 June 2019
9. Learning Design system homepage. http://learning-design.sch.gr. Accessed 9 June 2019
10. Márquez-Vera, C., Cano, A., Romero, C., Noaman, A.Y.M., Mousa Fardoun, H., Ventura, S.:
Early dropout prediction using data mining: a case study with high school students. Expert
Syst. 33(1), 107–124 (2016)
11. Megalou, E., Kaklamanis, C.: Photodentro LOR, the Greek national learning object repository.
In: INTED2014 Proceedings, pp. 309–319 (2014)
12. “MySchool” homepage. https://myschool.sch.gr/. Accessed 9 June 2019
13. Nikolaou, S.M., Papa, M., Gogou, L.: Early school leaving in Greece and Europe and educa-
tional inequality: actions and policies against educational and social exclusion. Eur. J. Soc.
Sci. Educ. Res. 5(1), 212–220 (2018)
14. Pardo, A., Siemens, G.: Ethical and privacy principles for learning analytics. Br. J. Edu.
Technol. 45(3), 438–450 (2014)
A Case Study on How Greek Teachers Make Use of Big Data Analytics 9
1 Introduction
Reading assessment and mathematics assessment have always been as the major sub-
jects in Taiwan Assessment of Student Achievement (TASA) since 2004 as well as in
the PISA (Programme for International Student Assessment) report. More important,
reading comprehension cannot be just limited to the language field. It should be seen as
an important skill in all the fields throughout the school years of the elementary school.
When it comes to mathematics, many primary school students fail to solve mathematics
problems, which is not because of their poor computing ability, but of their failure of
understanding the problem, meaning that those students are unable to succeed in the first
step of solving the problem (Huang 1996). As indicated, problem-solving is the core of
mathematics curriculum. Thus, students need to develop reading comprehension ability
to solve the mathematics problems.
2 Literature Review
(1) Understand the problem: The problem solver must understand the narrative of the
problem and point out the main idea of the problem, meaning that the problem
solver must characterize the problem.
(2) Devise a plan: The problem solver must find out the relationship between the known
and the unknown, and set up the procedures to deal with the problem.
(3) Carry out the plan: The problem solver must perform various calculations and other
necessary operations.
12 C. Tung et al.
(4) Review/extend: The problem solver has to reexamine the process carefully, trying
to see how this experience can help solve other problems.
In this study, the teacher applied the teach-back method, the students were paired
(one high achievement student with one low achievement student). They introduce their
understanding of the topic to each other. Then, tell each other the content of the study
and question the content. If they have different opinions, intervention of teachers will
occur. If they reach the consensus, the student will sum up.
Tell each other the content of the study and question the content
To sum up
3 Method
The Subjects
The study was conducted in two grade-six classes in an elementary school in Taiwan;
each class was composed of 24 students. The 48 students all have studied the basic
concepts of numbers and quantities and algebra. The units discussed in this paper were
selected from the students’ mathematics textbook, including algebra topics mainly on
calculating the numbers of chickens and rabbits in a same cage with the information
of their feet. The two classes were randomly selected as the experimental group or the
control group.
In the experimental group, four students were in a group. The heterogeneous grouping
strategy, which means the group consists of low and high achievers was used. Each group
was then again grouped into two pairs for the teach-back activity. The teach-back method
was implemented for students to solve the math problem. After each group reached
consensus, the group members used the tablet to show how to solve the math problem,
using the HiTeach interactive learning platform to share their problem-solving steps.
In the control group: Four students were in a group. The heterogeneous grouping
strategy was used for grouping. The group members used the tablet to show how to solve
the math problem, also using the HiTeach interactive learning platform to share their
problem-solving steps in the traditional cooperative learning method.
Fig. 2. HiTeach interactive learning platform: a screen showing the math problem
14 C. Tung et al.
The Instruments
The research data came from students’ test papers, a learning-attitude questionnaire,
students’ worksheets, and the data on the HiTeach interactive learning platform.
The math test papers of this study were designed by the teachers. The test papers
were also validated by three senior math teachers and two math specialties. The expert
validity as well as the content validity was achieved. The pre-test and post-test of student
math performance were obtained from the test papers. The difficulty of the test paper is
at 0.5, which is a mean value of the data contained in Table 1 for the level of difficulty.
After the pilot test, the 18 official test items were selected. The difficulty distribution
table of the test items is shown in Table 1.
The learning attitude questionnaire was designed to understand how learning atti-
tudes could be influenced by the learning activities—including the teach-back method
as well as the interactive teaching system. It was based on a five-point Likert scale, in
which 5 indicates the strongest degree of agreement while 1 indicates the weakest degree
of agreement. Ten items from the questionnaire were selected for the discussion in this
paper. Items 1 to 5 were on the aspect of cooperative learning and items 6 to 10 were on
the aspect of students’ learning process
Data Collection
The study was carried out within four weeks. The pre-test was implemented in the first
week. During the second and the third week, the implementation of teach-back method
unit was carried out in the experiment group while in the control group the traditional
Applying the Teach-Back Method and Mobile Technology 15
discussion strategy was implemented. In the fourth week, the post-test and the attitude
questionnaire were conducted. The teaching process of week 2 and week 3 is as follows:
(1) The whole class was taught: the teacher illustrated the example in the textbook.
(2) Understanding of the problem: The teacher wrote down similar questions, and
the students wrote down their understanding of the questions and their process of
solving the math problems. Students did not discuss what they wrote.
(3) Sharing understanding: The result of the second step was put into discussion. The
experimental group members played the role of the commentator in turn, explaining
the meaning of the question and the way to solve the math problem. If the methods
of solving the problem from both parties were similar, the consensus was reached.
In the control group, as four members were in each group, adopting a traditional
group discussion strategy. If the answers within the group were similar, then the
consensus was reached. If the answers were not the same, group members would
continue discussing for the consensus answer.
Table 2. Comparison between the experimental group and the control group
16 C. Tung et al.
(4) Sharing between groups: Each group uploaded a snapshot of their answers to the
interactive teaching system, and the learners could refer to other groups’ results and
reviewed the answer all together.
(5) Summary: Finally, the teacher summarized and reviewed the main point of the math
activity (Table 2).
4 Research Results
The Pre-test Results
In order to understand whether there are significant differences between the two classes
of students in the concept of “how to solve the problem” before the experiment, the
experiment group and the control group were tested by the pre-test, and the independent
sample t-test was analyzed by SPSS statistics software for the purpose of understanding
the difference between the two groups of students. The results are shown in Table 3:
The average scores of the experiment group and the control group were 35.71 and
42.48, respectively. The Levene test of the homogeneity of the variance was of little
significance (F = 2.88, p = .096 > .05), indicating there was no significant difference
between the experimental group and the control group before the experiment, that is, the
mathematics ability of the two classes before the experiment was equivalent.
Table 5. Mann-Whitney test for post-test scores of high and low achievement students
After the implementation of the HiTeach interactive teaching platform with the teach-
back method, the low-achieving students in the experimental group not only improved
the average score, but also achieved significant differences in the learning outcomes,
showing the low achievement students in the experimental group performed better than
the low achievement students in the control group.
Cooperative Learning
The independent sample t-test was performed to test the influence of the treatment on the
students’ learning attitudes. In the cooperative learning phase, Items 1 to 5 were used to
access students’ behaviors and opinions during cooperative learning in the experiment
group and the control group. The results are shown in Table 6.
18 C. Tung et al.
The results of Item 1, 4, and 5 show that the experiment group students are signifi-
cantly different from the control group. The researchers found that the feedback method
made the students in the experiment group more willing to put forward their own opin-
ions when faced with differences of opinions or when they had doubts about the other
students’ explanation. The degree of cooperation within the group was also better than
that of the control group.
Thinking Process
Items 6 to 10 are used to access students’ opinions regarding using the worksheet during
the group discussion session. The results are shown in Table 7. It can be seen that students
in the experiment group (using the Teach-back method) responded with better results
than the students in the control group (the traditional cooperative learning). The reason
is that the members of the Teach-back method group have more interactions within
the group and also more interactions between the groups, which makes students have
more opportunities to present their own thinking process, and when they listen to others’
solutions, they can also recognize and overcome their weaknesses.
Applying the Teach-Back Method and Mobile Technology 19
5 Conclusion
The study explored elementary students’ mathematics learning efficacy, using the teach-
back method, and their attitudes, using the tablet and the HiTeach interactive learning
platform. The results show that both the experiment group of “HiTeach interactive teach-
ing platform integrated with the teach-back method” and the control group of “HiTeach
interactive teaching platform integrated into traditional cooperative learning teaching”
can effectively improve the effectiveness of students’ ability of solving math problems.
The learning achievement of the low-achieving students in the experiment group was
significantly better than that of the low-achieving students in the control group. More-
over, students show much favor to the incorporation of technology into the teach-back
method in a mathematics classroom, thinking that in this way, they can show their under-
standing of the meaning of questions by sharing their own ideas. Through the sharing of
the HiTeach learning platform, they can also view the way other groups solve questions.
As a result they can learn more different ways of thinking and, in the meantime, improve
their motivation and interests in learning mathematics.
References
Gillies, R.M.: Cooperative learning: review of research and practice. Aust. J. Teach. Educ. 41
(2016). https://doi.org/10.14221/ajte.2016v41n3.3
Huang, Z.: Mathematical Teaching Method. Normal University Book Company, Taipei (1996)
Johnson, D.W., Johnson, R.T., Holubec, E.J.: Cooperation in the Classroom, 9th edn. Interaction
Book Company, Edina (2013)
Polya, G.: How to Solve It: A New Aspect of Mathematical Method. Princeton University Press,
Princeton (1945)
20 C. Tung et al.
Sharples, M., et al.: Innovating Pedagogy 2016: Open University Innovation Report 5. The Open
University, Milton Keynes (2016)
Slavin, R.E.: Cooperative learning and academic achievement: why does groupwork work? Anales
de Psicología/Ann. Psychol. 30, 785–791 (2014). https://doi.org/10.6018/analesps.30.3.201201
Wakefield, D.V.: Math as a second language. Educ. Forum 64, 272–279 (2000)
Taiwan Achievement of Student Achievement (2004). https://www.naer.edu.tw/files/11-1000-
1408-1.php?Lang=zh-tw
EFL Writing Assessment: Peer Assessment vs.
Automated Essay Scoring
Abstract. This study aimed to explore problems and potentials of new technolo-
gies in English as foreign language (EFL) writing education. Forty-six students
as a foreign language (EFL) learners in a Chinese university participated in this
study. They submitted their draft to Pigai Network and Scholar Network separately
and received automated essay scoring (AES) and peer assessment (PA) feedback.
Results showed a moderate, positive partial correlation between PA and AES,
controlling for performance level. The EFL learners in China preferred AES over
PA. These findings raise several relevant issues in how to improve peer assessment
feedback effectively, such as writing rubric in peer assessment, specialized peer
assessment tool, technology assistant and peer feedback.
1 Introduction
In the past few decades, several forms of assessment have been used in English as a
foreign language (EFL) classes, including teacher assessment, peer assessment (PA)
and automated essay scoring (AES). According to the effectiveness of assessment, peer
assessment and teacher feedback are widely used in writing classes [1, 2]. As technology
advances, independent research on AES has only recently emerged [3–5] and has played
an important role on writing teaching.
However, teacher assessment is regarded as the most authoritative way of assessment.
Many teacher report that correction of student essays consumes the largest proportion of
their time: about 40 min to comment on each individual essay times 3 essays per student
tines 46 students per class yields 5, 5620 min or 92 h [6]. One alternative is having peers
evaluate the quality or academic performance of each student (peer assessment, PA). Past
studies show mixed results regarding the effectiveness of peer assessment for improving
EFL, possibly because of its validity [7, 8] and the variability of assessment tasks [9]. For
example, Chang et al. indicated that peer assessment based on web-based portfolio failed
to show students’ performance and was not a reliable and valid assessment method [7].
Automated essay scoring (AES) might address both issues, and students’ perceptions
towards PA and AES has an important role in shaping students’ views and behaviors in
English writing, so both assessment methods are evaluated in this study.
Forty-six college students participated in this study. Two essays written by the stu-
dents were submitted and implemented peer assessment and automated essay scoring.
Quantitative and qualitative data were examined to yield findings, followed by a detailed
discussion.
2 Related Work
Significant advantages of PA have been reported in the literature [10, 11]. Topping et al.
reported that PA showed adequate reliability and validity in many kinds of applica-
tions, and the PA effects were as good as or better than teacher assessment effects [12].
Meek et al. also showed that higher participation in the peer review task correlated with
higher performance [13]. However, some researchers found that PA scores and teacher
assessment score differed significantly [14, 15] and suspected the validity of the PA.
In terms of perception towards PA, some students believe that they can gain benefit
from PA [13]. However, some students found PA difficult, uncomfortable and time-
consuming [16]. In spite of previous advantages and disadvantages of PA and mixed
attitudes towards PA, there are also some potential limitations to be explored next.
First of all, whether PA in EFL writing are the same as teacher assessment in China.
Additionally, the effectiveness of PA adopted in writing assessment is remained more
profound research as most of the related work focused on.
Automated essay scoring (AES) can supplement writing assessment, but most studies
of its validity show mixed findings. Fang showed that a majority of the students bene-
fited from AES and that the automated feedback had a positive effect on writing skill
development, particularly on the form rather than content [17]. Human raters and AES
showed highly consistency and some evidence of validity supported AES [18].
However, how students perceived the effectiveness in EFL writing were mixed.
Chen and Cheng found that students favoured using AES for early drafting and revising
[19], and showed writing improvement with respect to accuracy and learner autonomy
awareness [20]. However, another study showed that students’ perceptions had minimal
impact on their AES use to write and revise successfully, though some students continued
to use it or would recommend it to friends [21].
in Taiwan generally opted for PA over AES [22]. Inspired by previous ideas, the present
study adopted both PA and AES to examine the difference between PA and AES, and
explore the students’ perceptions in EFL writing.
3 Research Design
This study explored the differences between PA and AES in an EFL writing learning
environment and determined students’ perceptions toward the two assessments methods.
Quantitative data and qualitative questionnaire data were collected from 46 college
students, along with 92 essays’ PA score and AES score to ascertain the suitability of
PA or AES as an assessment tools for EFL students. As explained in previous studies,
the effectiveness of AES and PA heavily depends on how the assessment is conducted
because the method can affect the validity of the feedback. Specifically, this study poses
the following research questions:
3.1 Participants
3.2 Instruments
Three network tools were adopted in this study in EFL writing assessment for PA and
AES. Pigai Network is a Chinese AES network tool often used in EFL writing assess-
ment in this university. Based on corpus and cloud computing technology, Pigai pro-
vides an automatic online correction service for English essays, along with scores and
comments about the composition in real time. Scholar Network, an online academic
information service platform, provides academic information management, academic
exchange services, and a teaching curriculum platform to support an academic commu-
nity. Questionnaire Star, a professional online questionnaire survey platform, provided
an online questionnaires and analysis.
24 M. Lu et al.
78.07 and 80.61 vs. 66.74). Also, the standard deviations of PA (11.70 and 8.03) far
exceeded those of AES (6.71 and 4.79) and the teacher (6.34 and 6.34). After teacher
intervention for the second PA, the standard deviations of its scores fell from 11.70 to
8.03.
Moreover, the results showed that there was no significant correlation between PA
and student writing performance. But the teacher intervention can effectively improve
the correlation coefficient between PA and writing performance. Also, the same results
showed that there was moderate, positive partial correlation between AES and writing
performance after teacher intervention (r (43) = 0.26, Sig. = .08 in the first essay, and r
(43) = 0.32, Sig. = .02 in the second). In summary, teacher intervention can effectively
improve the correlation coefficient of between PA and AES and writing performance.
26 M. Lu et al.
Table 3 indicated that the students’ attitudes toward PA and AES were often similar
based on paired t-tests.
Items PA(AES)
Mean SD t-value n Sig.
RQ1 3.36 (3.96) 0.86 (0.74) 4.29 44 0
RQ2 3.36 (3.91) 0.86 (0.79) 4.43 44 0
RQ3 3.53 (3.89) 0.66 (0.71) 4.18 44 0
RQ4 3.56 (3.53) 0.76 (0.79) −0.21 44 0.84
RQ5 3.70 (3.51) 0.70 (0.79) −1.43 44 0.16
RQ6 3.42 (3.47) 0.66 (0.76) 0.42 44 0.67
RQ7 3.56 (3.62) 0.79 (0.65) 0.62 44 0.54
RQ8 3.38 (3.42) 0.61 (0.84) 0.4 44 0.69
RQ9 3.38 (3.58) 0.78 (0.75) 1.32 44 0.19
RQ10 3.58 (3.51) 0.75 (0.79) −0.55 44 0.58
According to the first question “I regard Pigai (vs. my classmates) as the real audi-
ence”, the results (Table 3) showed that the students showed more positive attitudes
toward Pigai (Mean = 3.96, SD = 0.74) than PA (Mean = 3.36, SD = 0.86). Also,
for questions “I highly value the comments from Pigai (vs. my classmates’ comments)
on my writing” and “I adopt comments from Pigai (vs. my classmates’ comments) for
revision,” students valued Pigai comments more than from classmates’ comments.
Student responses to three open-ended questions shed more light on these results.
Responses to the first question “What are the factor influences the implementation of
PA?” included emotional factors, writing rubric, performance level, peer relationships,
PA tools, technology and the relationship between classmates.
The second question asked “Which tool is more suitable for PA?” Almost all the
students believed that PA requires the support of online tools and one-third of students
view a special-purpose online PA tool is necessary. Also, more than 90% of the students
in the class take a view that instant feedback is more useful during essay revision
The third question asked “What are the advantages and disadvantages of implement-
ing PA in classroom?” PA advantages included: “reduce the burden on teachers, pro-
vide instant feedback, improve learning and communication experience, share different
writing ideas, help self-assessment and understand own shortcomings more comprehen-
sively, and recognize some problems so that you can avoid writing your own text in the
future.” Disadvantages were: “lack of rating scale, unbalanced performance level, a bit
long time and not timely enough, not know how to modify the essay and not confident
to evaluate the classmates’ essay”.
EFL Writing Assessment: Peer Assessment vs. Automated Essay Scoring 27
5.2 Limitations
First, the sample size is small and the qualitative and quantitative data only reflects the
learners who participated this study. Therefore, this study is more of a nature of exper-
imental or exploratory research, and can’t make strong statistical claims. Second, there
were no specialized PA tools, and Scholar Network cannot provide instant peer feedback
or manage the detailed PA process. And it can’t provide high quality peer assessment
conditions. Third, we did not explore the effectiveness of the two assessment meth-
ods regarding the content quality of the essay. Meanwhile, AES can provide immediate
holistic and analytical feedback [25, 26].
Funding. This work was supported by [Teaching Quality and Teaching Reform Project in Guang-
dong Province] under Grant [number 236: No. 201, No. 218]; [Guangdong Provincial Philosophy
and Social Sciences Project] under Grant [number GD18WXZ18]; and [The Ministry of Educa-
tion’s Higher Education Department, the second batch of industry-university collaborative edu-
cation project] under Grant [number 201802083033]; [Guangdong University of Foreign Studies
Postgraduate International Talents Training Innovation Project].
28 M. Lu et al.
Appendix
Perceptions Toward Feedback
References
1. Berg, E.C.: The effects of trained peer response on ESL students’ revision types and writing
quality. J. Second Lang. Writ. 8(3), 215–241 (1999)
2. Yang, M., Badger, R., Yu, Z.: A comparative study of peer and teacher feedback in a Chinese
EFL writing class. J. Second Lang. Writ. 15(3), 179–200 (2006)
3. Attali, Y., Burstein, J.: Automated essay scoring with e-rater®; V.2.0. J. Technol. Learn.
Assess. 4(2), i–21 (2006)
4. Dikli, S.: Automated essay scoring. Turk. Online J. Distance Educ. 7(1), 735–738 (2006)
5. Deane, P.: On the relation between automated-essay scoring and modern views of the writing
construct. Assessing Writ. 18(1), 7–24 (2013)
6. Sommers, N.: Responding to student writing. Coll. Compos. Commun. 33(2), 148–156 (1982)
7. Chang, C.-C., et al.: Reliability and validity of web-based portfolio peer assessment: a case
study for a senior high school’s students taking computer course.”. Comput. Educ. 57(1),
1306–1316 (2011)
EFL Writing Assessment: Peer Assessment vs. Automated Essay Scoring 29
8. Chang, C.-C., Yan, C.-F., Tseng, J.-S.: Perceived convenience in an extended technology
acceptance model: mobile technology and English learning for college students. Australas. J.
Educ. Technol. 28(5), 809–826 (2012)
9. Topping, K.J.: Methodological quandaries in studying process and outcomes in peer
assessment. Learn. Instr. 20(4), 339–343 (2010)
10. Falchikov, N.: Product comparisons and process benefits of collaborative peer group and
self-assessments. Assess. Eval. High. Educ, 11(2), 146–166 (1986)
11. Falchikov, N., Goldfinch, J.: Student peer assessment in higher education: a meta-analysis
comparing peer and teacher marks. Rev. Educ. Res. 70(3), 287–322 (2000)
12. Topping, K.: Peer assessment between students in colleges and universities. Rev. Educ. Res.
68(3), 249–276 (1998)
13. Meek, S.E.M., Blakemore, L., Marks, L.: Is peer review an appropriate form of assessment
in a MOOC? Student participation and performance in formative peer review. Assess. Eval.
High. Educ. 42(6), 1000–1013 (2017)
14. Hovardas, T., Tsivitanidou, O.E., Zacharia, Z.C.: Peer versus expert feedback: an investigation
of the quality of peer feedback among secondary school students. Comput. Educ. 71, 133–152
(2014)
15. Tsai, C.-C., Lin, S.S.J., Yuan, S.-M.: Developing science activities through a networked peer
assessment system. Comput. Educ. 38(1–3), 241–252 (2002)
16. Hanrahan, S.J., Isaacs, G.: Assessing self-and peer-assessment: the students’ views. High.
Educ. Res. Dev. 20(1), 53–70 (2001)
17. Fang, Y.: Perceptions of the computer-assisted writing program among EFL college learners.
Educ. Technol. Soc. 13(3), 246–256 (2010)
18. Enright, M.K., Quinlan, T.: Complementing human judgment of essays written by English
language learners with e-rater® scoring. Lang. Test. 27(3), 317–334 (2010)
19. Chen, C.-F.E., Cheng, W.-Y.E.C.: Beyond the design of automated writing evaluation: peda-
gogical practices and perceived learning effectiveness in EFL writing classes. Lang. Learn.
Technol. 12(2), 94–112 (2008)
20. Wang, Y.-J., Shang, H.-F., Briody, P.: Exploring the impact of using automated writing eval-
uation in English as a foreign language university students’ writing. Comput. Assist. Lang.
Learn. 26(3), 234–257 (2013)
21. Roscoe, R.D., et al.: Presentation, expectations, and experience: Sources of student percep-
tions of automated writing evaluation. Comput. Hum. Behav. 70, 207–221 (2017)
22. Lai, Y.-h.: Which do students prefer to evaluate their essays: peers or computer program. Br.
J. Edu. Technol. 41(3), 432–454 (2010)
23. Jones, I., Wheadon, C.: Peer assessment using comparative and absolute judgement. Stud.
Educ. Eval. 47, 93–101 (2015)
24. Yu, F.-Y., Wu, C.-P.: Different identity revelation modes in an online peer-assessment learn-
ing environment: effects on perceptions toward assessors, classroom climate and learning
activities. Comput. Educ. 57(3), 2167–2177 (2011)
25. Hoon, T.: Online automated essay assessment: potentials for writing development (2010).
Accessed 9 August 2006
26. Yeh, Y.-L., Liou, H.-C., Yu, Y.-T.: The influence of automatic essay evaluation and bilingual
concordancing on EFL students. English Teach. Learn. 31(1), 117–160 (2007)
Dimensions of Learning Organization
in Relation to Learning Time – Cross-Sectional
Study at Secondary Schools from the Czech
Republic
Vaclav Zubr(B)
1 Introduction
In many organizations, we encounter the human resource development model, where
education is a separate function. If we talk about a learning organization concept, learning
in the organization is supported [1]. Employees in a learning organization are expected to
learn, plan future skills, actions and risks and deal with issues, the learning organization
concept is supported e.g. by teamwork, sharing information and learning communities
[2, 3]. The core activity of the learning organization is organizational learning. Organiza-
tional learning can be characterized as a complicated, unplanned process that is alert and
effective, interactive and dynamic, continuous and persistent, developing and growing
and influenced by the knowledge base or cultural resources [4]. A part of organizational
learning could be personalized and adaptive learning.
In the Czech Republic, there were 1,297 high schools providing day-to-day educa-
tion in the school year 2017/2018, where 38,115 teachers were employed [5]. Schools
2 Methodology
3 Results
Altogether 1,304 secondary schools were addressed in the Czech Republic. In total 121
respondents participated in the study (9.28%).
The study was attended in total by 45 (37.19%) men and 76 (62.81%) women aged
25 to 70 years. A total of 115 respondents had university education, 4 respondents had
secondary education, 1 respondent had higher vocational education and 1 respondent
had primary education. Only 1 school represented an organization with more than 250
employees, the schools with fewer than 50 employees (a total of 73 schools) and schools
employing up to 250 employees (47 schools) were represented the most. If we evaluate
representation of schools by their type, secondary vocational schools (55.37%), grammar
schools (36.36%) and secondary vocational practice schools (8.26%) were represented.
Altogether, the study was attended by 87 (71.90%) executives and 34 (28.10%) tertiary
staff. The majority were respondents who worked in the organization up to 10 years
(39 respondents), followed by 11–20 years (33 respondents). On average, employees in
organizations work 18.32 years, the shortest period of practice was 0.5 years, the longest
was 49 years.
When comparing senior executives and executives, the senior executives are more
involved in education in relation to employment. There is no statistically significant
difference between common and senior employees (p = 0.132, α = 0.05, t-test). If we
compare employees from grammar schools and other secondary schools, there is no
statistically significant difference (p = 0.415, α = 0.05, t-test). At the same time, there is
no statistically significant difference in the significance level 0.05 between employees of
organizations with fewer than 50 employees and employees of organizations with 51 to
250 employees (p = 0.495). There is no statistically significant difference in relation to
the time of education between employees who work in the organization for less than 10
years and employees who work in the organization for more than 10 years (p = 0.326)
(Table 1).
significant difference between the respondents who spend 1–10 h a month with learning
and the respondents who learn for more than 36 h a month (p = 0.049) at the significance
level of 0.05. At the same time, there was a statistically significant difference found via
the t-test between respondents who learn for 21–35 h a month and respondents who learn
for more than 36 h a month (p = 0.012).
4 Discussion
Many authors throughout the world [11–17] deal with introducing the learning organi-
zation concept in organizations of different types. In many cases, these authors use the
DLOQ questionnaire to measure the concept of a learning organization [12–19]. Since
no further studies with DLOQ were carried out in the Czech Republic, this study’s aim
was to conduct a questionnaire survey at secondary schools in the Czech Republic and
evaluate the time devoted to learning in these organizations.
A total of 121 respondents from Czech Republic secondary schools were included in
this study. Although the return on questionnaires was relatively low (9.28%), the number
of respondents was comparable to other studies [15, 16]. Overall, 62.81% of women and
37.19% of men participated in the study. Percentage representation of women and men in
the study in the Czech Republic corresponds directly to the general share of women and
men in education, where according to the OECD the share of women in the secondary
degree of education is 63% [20].
Most respondents were 51–60 years old (47.11%) and 41–50 years old (23.97%).
This corresponds to the survey results of the Czech School Inspectorate of 2016, whereby
almost 70% of the principals of the studied secondary schools are over 50 years of age
and the average age of the teachers in the schools is 46.5 years [21].
The respondents who work in the organization for less than 10 years (32.23%)
have the largest representation. One could expect that these respondents will be mainly
young employees coming to work after school and those who are looking for a suitable
job for them at a younger age. Looking closer at the data, it is interesting that these are
respondents of different age groups (from the age of 25 to 70). This result may point to
a possible large fluctuation in education staff.
Teachers are expected to devote more time to education than headteachers (due
to a greater share of frontal teaching). It was found that most respondents (33.06%)
devote 11–20 h to education per month. Through the t-test, a statistically significant
difference was not found in relation to education at the significance level of 0.05 between
the position in the job, the type of organization and the time of employment in the
organization. The personalized and adaptive learning of teachers could be facilitated by
using learning phone applications or through online webinars.
When assessing the dimensions of a learning organization in relation to the time
spent on education, there is a statistically significant difference between respondents
who spend 1–10 h a month or 21–35 h a month and respondents who learn for more than
36 h a month. Compared to the same study from the Czech Republic conducted in the
IT sector, the results are higher in many dimensions [22].
A relatively small number of respondents who participated in the study may be
considered as limiting the study. In the future, it would be appropriate to carry out the
same study on a larger number of respondents.
34 V. Zubr
5 Conclusion
This study focused mainly on the learning time analysis at secondary schools in the
Czech Republic. The education results in relation to employment seem satisfactory in
respondents in education in the Czech Republic. The interdependence between learning
time and dimensions assessment of learning organization was demonstrated by t-test. In
this way, it can be appreciated that the period of education has a positive impact on the
assessment of the individual dimensions of the learning organization (and therefore the
development of the organization as a learning organization).
Although the education of employees in secondary schools in the Czech Republic
has been assessed as satisfactory, it would be beneficial to access new learning methods
for supporting the learning of employees. Also, it is necessary to carry out similar studies
in other sectors in the future to better compare the results and to design a more complex
solution applicable to more than one discipline.
Acknowledgement. The paper was written with the support of the specific project 2019 grant
“Determinants of Cognitive Processes Impacting the Work Performance” granted by the University
of Hradec Kralove, Czech Republic and thanks to help of students František Hašek and Jan
Petružálek.
References
1. Birdthistle, N.: Family Businesses and the Learning Organisation: A Guide to Transforming
the Family Business into a Learning Organisation. VDM Verlag Dr. Müller Aktiengesellschaft
& Co (2009)
2. Marquardt, M.J.: Building the Learning Organization. Davies-Black Publishing, Palo Alto
(2002)
3. Zubr, V., Mohelska, H., Sokolova, M.: Factors with positive and negative impact on learning
organization. In: Jedlicka, P., Maresova, P., Soukal, I. (eds.) Hradec Economic Days 2017,
Double-Blind Peer-Reviewed Proceedings of the International Scientific Conference Hradec
Economic Days 2017, vol. 7, no. 1, pp. 980–985. Gaudeamus, Hradec Kralove (2017)
4. Saadat, V., Saadat, Z.: Organizational learning as a key role of organizational success. Procedia
Soc. Behav. Sci. 230, 219–225 (2016)
5. Czech Statistical Office. https://www.czso.cz/documents/10180/61508174/2300421845.pdf/
c5380c95-eb6d-4160-8ec3-920deaf76b00?version=1.2. Accessed 28 May 2019
6. Kools, M., Stoll, L.: What Makes a School a Learning Organisation? OECD Publishing, Paris
(2016)
7. Watkins, K.E., Marsick, V.J.: Sculpting the Learning Organization: Lessons in the Art and
Science of Systematic Change. Jossey-Bass, San Francisco (1993)
8. Watkins, K.E., O’Neil, J.: The dimensions of the learning organization questionnaire (the
DLOQ): a nontechnical manual. Adv. Dev. Hum. Res. 15(2), 133–147 (2013)
9. Zubr, V.: Dimension of a learning organisation in the IT sector in the Czech Republic – case
study. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R. (eds.) SETE 2018. LNCS, vol.
11284, pp. 107–116. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03580-8_12
10. Institute for Digital Research and Education. https://stats.idre.ucla.edu/spss/faq/what-does-
cronbachs-alpha-mean/. Accessed 1 June 2019
Dimensions of Learning Organization in Relation to Learning Time 35
11. Marsick, V., Watkins, K.: Demonstrating the value of an organization’s learning culture: the
dimensions of the learning organization questionnaire. Adv. Dev. Hum. Res. 5(2), 132–151
(2003)
12. Cierna, H., Sujova, E., Habek, P., Horska, E., Kapsdorferova, Z.: Learning organization at
higher education institutions in the EU: proposal for implementing philosophy of learning
organization – results from research. Qual. Quant. 51, 1305–1320 (2017)
13. Reese, S.: Unlearning and the learning organization: revisited and expanded. Learn. Organ.
25(3), 210–212 (2018)
14. Palos, R., Stancovici, V.V.: Learning in organization. Learn. Organ. 23(1), 2–22 (2016)
15. Abo Al Ola, L.M.: The availability of Dimensions of Learning Organizations Questionnaire
(DLOQ) in educational college at Taif University from the perspective of employees. J. Educ.
Psychol. Sci. 18(1), 447–487 (2017)
16. Nazari, K., Pihie, Z.: Assessing learning organization dimensions and demographic factors
in technical and vocational colleges in Iran. Int. J. Bus. Soc. Sci. 3(3), 210–219 (2012)
17. Voolaid, K., Ehrlich, Ü.: Organizational learning of higher education institutions: the case of
Estonia. Learn. Organ. 24(5), 340–354 (2017)
18. Mohamed Ali, M., Abdulhamed, K.R., Aljamoudi, S.S.: The dimensions of the learning
organization in Omani school from employees point of view. Int. J. Sci. Technol. Res. 4(11),
267–271 (2015)
19. Park, J.H.: Validation of Senge’s learning organization model with teachers of vocational high
schools at the Seoul Megalopolis. Asia Pac. Educ. Rev. 9(3), 270–284 (2008)
20. OECD. https://www.oecd-ilibrary.org/docserver/54f0ef95-en.pdf?expires=1560189519&
id=id&accname=guest&checksum=7D33376875CDF7E59C3EE6B48B05A4CE. Accessed
29 May 2019
21. Czech School Inspection. http://www.csicr.cz/html/VZCSI2015_2016/html5/index.html?&
locale=CSY. Accessed 3 June 2019
22. Zubr, V., Mohelska, H.: Learning time analysis – case study in the IT sector in the Czech
Republic. In: Al-Sharhan, A., et al. (eds.) I3E: Conference on e-Business, e-Services and
e-Society 2018, LNCS, vol. 11195, pp. 21–29. Springer, Heidelberg (2018). https://doi.org/
10.1007/978-3-030-02131-3_3
How Much Is Online Community Engaged
in Learning Content? Case of World Top
Universities’ Facebook
Pavel Bachmann(B)
University of Hradec Kralove, Rokitanskeho 62, 50003 Hradec Kralove, Czech Republic
pavel.bachmann@uhk.cz
Abstract. Social media (SM) has become a part of everyday life, including higher
education and its learning environment. Therefore, this study aims on understand-
ing of content published on SM, its learning impact, as well as on identification of
relevant level of community engagement. Research sample used official Facebook
sites of six world top universities, specifically posts (N = 120) communicated dur-
ing November 2018 were investigated in detail. A new scale, specifically designed
for this study, was designed to capture a learning potential of content communi-
cated on the SM. An analysis oriented on main characteristics of engagement: its
responsiveness measured by number of “likes” and other available symbols, its
involvement measured by number of comments, its virality, measured by number
of sharings, and finally its total engagement expressed as the sum of preceding
characteristics. Several interesting results are provided in the study: the research
is the most frequent topic communicated by top universities; different content
strategies are taken by universities to engage their communities; and different
engagement levels exist according to the university as well as according to the
content published on its Facebook site.
1 Introduction
Social media are not only technical tools [1]; they are rather collections of online com-
munities, people and human interactions. Communities that are ready to be engaged,
ready to discuss, and ready to learn and to be educated. There are numerous studies on
use of Social Media (SM) in higher education environment [2–5].
Despite of rather general acceptance of SM tools in academic environment and even
its implementation in learning processes [3, 6–9] some negative or more complex com-
ments on SM use in the field are discussed also. Manca and Ranieri [4, 10] in their recent
studies documented that there are still several barriers for use of SM in academic environ-
ment as cultural resistance, pedagogical issues, or institutional constraints. Application
of SM in this learning environment depends mainly on a field of discipline and the
personality of teacher. Moreover, in 2015 Alt [11] introduced a scale reflecting psycho-
logical behavior of undergraduate students including SM engagement, fear of missing
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 36–45, 2020.
https://doi.org/10.1007/978-3-030-38778-5_5
How Much Is Online Community Engaged in Learning Content? 37
out, and academic motivation. Sharma, Joshi, and Sharma [12] predicted determinants
that affect students’ intention towards academic use of Facebook. Results showed that
resource sharing is the most influential determinant in the decision to use Facebook in
higher education, followed by perceived usefulness, perceived enjoyment, collaboration
and social influence.
Nowadays, various research studies confirm a positive impact of SM use on the learn-
ing processes of higher education [8, 9, 13]. Sobaih, Moustafa, Ghandforoush, et al. [8]
stress out that SM could be an innovative and effective tool for teaching and learning.
Moreover, Neier and Zayer [5] demonstrated in their study an openness for use SM in
higher education and interactive and information motivation on the side of students. Sim-
ilarly, Bozanta and Mardikyan [6] underline that perceived ease of use is a predictor of
perceived usefulness which implies SM tools can be successfully used for educational
purposes. SM improves peer interaction and course engagement of students and also
students’ interaction with faculty members. Also, peer interaction and course engage-
ment have positive significant effect on collaborative learning. Megele [13] in her paper
examined ways how to enhance students’ engagement and learning through embed-
ding social media technologies into the academic curriculum as a learning and assessment
strategy and showed that such based action learning model expands the interrelational
dimensions of students’ learning; mainly increased the students’ engagement and the
depth and breadth of their learning.
Even with numerous quantities of research on investigation SM and learning, the
lack of studies aiming on the description of current level of student’ engagement with
existing SM used by universities, was found. As a research gap was identified and from
the context mentioned above, to identify an engagement of university community in
online collaborative learning, the following research questions are investigated:
1. What learning content and how much of this content is offered on Facebook sites of
universities?
2. Does Facebook community (a) responsiveness, or (b) involvement, or (c) virality,
and (d) total engagement differ in top world universities?
3. Does Facebook community (a) responsiveness, or (b) involvement, or (c) virality,
and (d) total engagement differ according to educational/non-educational content
type of posts?
investigated. Universities published these twenty posts during the period of 15 days, in
average. The quantity of posts was limited due to the high time-consuming coding of
the content.
The sampled universities characteristics varied in the number of full-time students
enrolled (the biggest Harvard with over 22,000 students vs the smallest Caltech with over
2,000 of students), but namely in size of fan’ and follower’ base, where the differences are
even bigger (Harvard with 5 mio of fans vs Caltech with 380,000 students). Therefore,
later, the size of fan base is taken into account, when presenting the results. The list of
universities with more detailed characteristics is available in Table 1.
Table 1. Facebook sites of top universities: engagement, fan’, and follower’ bases
In addition, the intentions of using Facebook site by the university were studied.
All sampled universities have placed a Facebook site link on their main webpage, three
of them (Oxford, Cambridge and Caltech) with “connect with us” challenge. Two uni-
versities stated their intention of site existence in the “about” section: Caltech institute
considers its site to be “a place to encourage curious, thoughtful dialogue about topics
related to scientific discovery and the Caltech experience”, Oxford site intent is to pub-
lish news, events, and admissions. The other universities used this section for presenting
an ethical code of conduct for visitors.
2.2 Method
Content analysis of published Facebook posts was chosen as the main research method
of the study. This kind of investigation is a part of Internet-mediated research defined
by Hewson [14]. Only one coder was used for collection of the data to avoid distortion
of results thanks to different way of coding. Moreover, a new construct for measuring
the nature of content published was developed. The construct design reflected both, the
marketing needs of higher institutions on one side, and the educational and learning
mission, on the other side. Classification of content types of posts is depicted in Fig. 1.
More precise explanation of construct details and definition of individual categories
and examples of posts is formulated and available in Table 2.
How Much Is Online Community Engaged in Learning Content? 39
Non- Research
educa- and offline
PR and tional Teachers, teaching Online
CSR Scholarships, lectures,
activities Internship workshops
Non- Learning
learning content
content
“wow”, “sad”, and “angry” were introduced besides the original “like”. New responses
on the post published enable us to better recognize the views and needs of the users.
The second metrics - involvement is expressed by the number of comments. Writing a
comment requires much higher participation, consideration and thinking of students (or
other users) than a simple one-click reaction. Virality, expressed through the number of
posts shared, is a metrics with a high impact on the expansion of communication across
the other sites. In this way, the communication can affect much broader auditorium than
just page fans and/or followers base. The last metrics, total engagement rate is than
sum of preceding three metrics. Although, in practice the weight of latter two metrics is
multiplied at least three times, due to its much higher real engagement, in this study and
for better understanding of results, the metrics will be only a single sum of the metrics
mentioned.
The statistical significance of differences in engagement characteristics according to
the university as well as the content communicated to the public was analyzed with the
use of MS Excel and one-way analysis of variance.
3 Results
3.1 Universities and Engagement
the highest total engagement was found in MIT (702 engaged users per post) and Stanford
(610). The medium engagement exists in Harvard and Caltech (nearly 400 users). The
lowest level of engagement was found in communities of British Oxford and Cambridge.
The results analysis confirmed statistically significant differences among universities in
both total engagement per post (p-value < 0.001) as well as total engagement per fan
base (p-value = 0.0009). Detailed results are available in Table 4.
Table 4. Facebook sites of top universities: engagement, fan’, and follower’ bases (N = 120)
(Ranking) Name Total engagement per post Total engagement per fan basea
(1) Univ. of Oxford 781 216
(2) Univ. of Cambridge 461 217
(3) Caltech 143 378
(3) Stanford Univ. 777 610
(5) MIT 796 702
(6) Harvard Univ. 2,112 397
a In mio of fans
Table 6. Content type and its responsiveness, involvement, and virality (N = 120)
Total Engagement
The level of total engagement complies with the levels already found for previous engage-
ment characteristics. The results analysis confirmed only weak statistical difference sig-
nificant on 0.1 level in total engagement according to the content of posts (p-value =
0.094). All the details are available in Table 7.
How Much Is Online Community Engaged in Learning Content? 43
Content of posts Number of posts Total engagement Total engagement per post
1- PR and information messages 35 32,218 895
2- Events (without learning ones) 6 5,356 1,071
3- Instructors, Fellowship, 15 10,491 699
Scholarship
4- Research findings, students’ 62 30,760 496
projects
5- Online teaching 2 1,889 945
In total 120 68,758 573
Obviously, there are some limitations of the study, which should be discussed. At
first, although the number of posts is enough for collection of some findings in the
field, its increase would make given values more trustworthy as well as allow more
44 P. Bachmann
References
1. King, P.: Social networks: getting your organization working for them! In: Geier Jr., P.H.,
Greenfield, J.M., Hart, T., MacLaughlin, S. (eds.) Internet Management for Nonprofits:
Strategies, Tools and Trade Secrets, p. XIII. Wiley, Hoboken (2010)
2. Cerna, M., Svobodova, L.: Current social media landscape. In: Conference Proceedings on
Efficiency and Responsibility in Education 2013, Prague, Czech Republic (2013)
3. Lam, J.-H., Ma, W.W.K.: When and how does learning satisfy? Working collaboratively online
with a clear purpose. Int. J. Innov. Learn. 23(4), 400–415 (2018)
4. Manca, S., Ranieri, M.: “Yes for sharing, no for teaching!”: social media in academic practices.
Internet High. Educ. 29, 63–74 (2016)
5. Neier, S., Zayer, L.T.: Students’ perceptions and experiences of social media in higher
education. J. Mark. Educ. 37(3), 133–143 (2015)
6. Bozanta, A., Mardikyan, S.: The effects of social media use on collaborative learning: a case
of Turkey. Turk. Online J. Distance Educ. 18(1), 96–110 (2017)
7. Choo, E.K., Ranney, M.L., Chan, T.M., et al.: Twitter as a tool for communication and knowl-
edge exchange in academic medicine: a guide for sceptics and novices. Med. Teach. 37(5),
411–416 (2015)
8. Sobaih, A.E., Moustafa, M.A., Ghandforoush, P., Khan, M.: To use or not to use? Social media
in higher education in developing countries. Comput. Hum. Behav. 58, 296–305 (2016)
9. Zheng, B.B., Niiya, M., Warschauer, M.: Wikis and collaborative learning in higher education.
Technol. Pedagog. Educ. 24(3), 357–374 (2015)
10. Manca, S., Ranieri, M.: Facebook and the others. Potentials and obstacles of social media for
teaching in higher education. Comput. Educ. 95, 216–230 (2016)
11. Alt, D.: College students’ academic motivation, media engagement and fear of missing out.
Comput. Hum. Behav. 49, 111–119 (2015)
12. Sharma, S.K., Joshi, A., Sharma, H.: A multi-analytical approach to predict the Facebook
usage in higher education. Comput. Hum. Behav. 55, 340–353 (2016). Part A
How Much Is Online Community Engaged in Learning Content? 45
13. Megele, C.: eABLE: embedding social media in academic curriculum as a learning and
assessment strategy to enhance students learning and e-professionalism. Innov. Educ. Teach.
Int. 52(4), 414–425 (2015)
14. Hewson, C.: Gathering data on the Internet. Qualitative approaches and possibilities for mixed
methods research. In: The Oxford Handbook of Internet Psychology. Oxford University Press,
Oxford (2007)
15. Bachmann, P.: Citizens’ engagement on regional governments’ Facebook sites. Empirical
research from Central Europe. In: Conference Proceedings of Hradec Economic Days 2019,
vol. 1, pp. 6–14. University of Hradec Kralove (2019)
HandLeVR: Action-Oriented Learning in a VR
Painting Simulator
Raphael Zender1(B) , Pia Sander2 , Matthias Weise1 , Miriam Mulders2 , Ulrike Lucke1 ,
and Michael Kerres2
1 University of Potsdam, August-Bebel-Str. 89, 14482 Potsdam, Germany
{raphael.zender,matthias.weise,ulrike.lucke}@uni-potsdam.de
2 Universität Duisburg-Essen, Universitätsstr. 2, 45117 Essen, Germany
{pia.sander,miriam.mulders,michael.kerres}@uni-due.de
1 Motivation
A competence–oriented approach in vocational education and training requires com-
prehensive, action-oriented learning units with learning progress checks. By observing
actions in authentic learning and exercise situations, it must be possible to become insight
into the underlying competencies, both through reflection (by the trainee) and evaluation
(by the trainer). Problems with the consistent implementation of this requirement are
common, for example, in the training of vehicle painters. Here, various techniques for
applying individual layers of paint to workpieces must be trained. Adequate, frequent
and action-oriented training, however, is hampered by economic, physical, and social
factors.
With Virtual Reality (VR), psychomotor coordination and skills can be trained as
discussed in explorative learning approaches. VR technology enables a high degree
of immersion and authenticity of the learning situation, allowing learners to immerse
themselves in a learning world where they can control their learning process to a high
degree and learn by exploring the digital artifacts. In addition, painting is predestined
for VR use. For example, no haptic feedback from the 3D workpieces is required (apart
from the paint spray gun), as these are not touched during paint application.
The aim of the HandLeVR project is to develop and evaluate an effective training
system with a central VR learning application - the VR Painting Simulator. The system
will be used for training and following evaluation of paint applications on 3D workpieces.
It consists of an authoring tool for trainers as well as the VR learning application and
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 46–51, 2020.
https://doi.org/10.1007/978-3-030-38778-5_6
HandLeVR: Action-Oriented Learning in a VR Painting Simulator 47
a reflection application for trainees. The instructional and technical conceptions as well
as intermediate results are presented in the following sections.
2 VR Painting Simulator
1 https://handlevr.de.
48 R. Zender et al.
painters ask for more individual learning paths that can be customized according to
trainees’ proficiency level. Therefore, both approaches will be compared:
A. Explorative approach: Learners have access to all VR learning tasks. They can freely
navigate through the learning application and independently decide which learning
task they perform.
B. Systematic approach: The trainer or educational institution gives a pre-defined learn-
ing path that the trainees have to work through. However, the learning path may be
customized according to the proficiency level of an individual trainee.
shop of a German automobile manufacturer were carried out with (1) the trainer for car
painting, (2) three trainees and (3) two former trainees now working on the production
site. In addition, a typical training session in the painting shop was observed. The 4C/ID
model served as a basis to design the interview scheme and the observation guidelines.
The collected data provided the input to design a first concept for the training of vehicle
painters in VR. In the second phase of the training analysis the developed concept was
validated with a group of trainers for vehicle painters working at a center for cross-
regional education for vehicle painters. The results of this second phase are currently
analyzed.
Independent from the instructional concept, a basic realistic representation of paint
jobs, the working environment and the tools used is required. Therefore, the early techni-
cal developments focused on both, an authentic VR setting in terms of a multifunctional
painting booth and paint gun as well as an accurate simulation of the basic painting pro-
cess. Figure 2 shows the current version of the painting booth including an exemplary
engine hood and the user interface (left image). The 3D models of the painting booth
and our highly authentic paint gun (right image) where created on basis of their real
counterparts. The user interface in the form of a monitor on the wall gives access to the
current functionalities of the application. Besides an engine hood, several parts of a car
were modelled in detail and integrated into the application.
The aim of the current prototype was realistic representation and behavior of the
spray cone and the paint job on the workpiece. To help the user find the correct distance
to the work-piece, a ray can be activated that turns green when the correct distance is
reached. Additional first in-process evaluation possibilities were integrated, allowing to
determine the quantity and costs of the paint used and how much of the paint has been
wasted. If too much paint is applied the paint runs down the workpiece. This is one of the
50 R. Zender et al.
Fig. 2. Paint booth with a blank car part and the user interface on the wall (left image) and the
current version of our highly authentic paint gun (right)
The VR Painting Simulator described in this article focuses on the promotion of action-
oriented learning of techniques for performing automotive painting work through virtual
reality. First, it is investigated which contribution VR technologies offer to enable action-
oriented learning in vocational education and training and how instructional design
principles can be applied in these technologies to ensure appropriate learning success.
The findings from this study will be transferred into a VR-supported training (the VR
Painting Simulator). This learning framework consists of an authoring tool for trainers
as well as a VR learning application and a reflection application for trainees.
Concerning the instructional design, the developed concept will be revised and val-
idated repeatedly. Therefore, a close collaboration with trainers and trainees in the field
of car painting as well as with computer scientists to transfer the instructional concept
into VR is needed. The resulting VR Painting Simulator and its prototypes are evaluated
continuously during the project as well as in dedicated field tests.
In addition to vehicle painters, the transferability of the results to other trades with
related learning activities (e.g. classical painters, welders) will be examined during the
project. The project team will transfer the project results to the vocational training centers
of the chambers of skilled trades located throughout Germany and provide appropriate
advice to accompany the practical application. In addition, the sustainability of the
project will be promoted by the final publication of the project results under an open
source license and as an open educational resource.
HandLeVR: Action-Oriented Learning in a VR Painting Simulator 51
References
1. Zender, R., Weise, M., von der Heyde, M., Söbke, H.: Lehren und Lernen mit VR und AR - Was
wird erwartet? Was funktioniert? In: Proceedings of DeLFI Workshops 2018 Co-located with
16th e-Learning Conference of the German Computer Society, CEUR Workshop Proceedings
(2018)
2. Mikropoulos, T.A., Natsis, A.: Educational virtual environments. A ten-year review of
empirical research (1999–2009). Comput. Educ. 56(3), 769–780 (2011)
3. Fowler, C.: Virtual reality and learning: where is the pedagogy? Br. J. Educ. Technol. 46(2),
412–422 (2015)
4. Kerres, M.: Wirkungen und Wirksamkeit neuer Medien in der Bildung. In: Education Quality
Forum. Wirkungen und Wirksamkeit neuer Medien, pp. 31–44. Waxmann, Münster (2003)
5. Hochberg, J., Vogel, C., Bastiaens, T.: Gestaltung und Erforschung eines Mixed-Reality-
Lernsystems. MedienPädagogik: Zeitschrift für Theorie und Praxis der Medienbildung 28,
140–146 (2017). Tagungsband: Bildung gemeinsam verändern: Diskussionsbeiträge und
Impulse aus Forschung und Praxis
6. Cheng, K.-H., Tsai, C.-C.: Affordances of augmented reality in science learning: suggestions
for future research. J. Sci. Educ. Technol. 22(4), 449–462 (2013)
7. Elliott, J.B., Gardner, M., Alrashidi, M.: Towards a framework for the design of mixed reality
immersive education spaces. In: 2nd European Immersive Education Summit-E-iED, pp. 63–
76 (2012)
8. Chen, P., Liu, X., Cheng, W., Huang, R.: A review of using Augmented Reality in Education
from 2011 to 2016. In: Popescu, E., et al. (eds.) Innovations in Smart Learning. LNET,
pp. 13–18. Springer, Singapore (2017). https://doi.org/10.1007/978-981-10-2419-1_2
9. van Merriënboer, J.J.G., Clark, R.E., de Croock, M.B.M.: Blueprints for complex learning:
the 4C/ID-model. Education Tech. Research Dev. 50, 39–61 (2002)
10. Hetemank, A., Mok, S.Y., CHU Research Group: Ist Lernsoftware wirklich effektiver, wenn
SchülerInnen den Lernprozess selbst in die Hand nehmen? (Kurzreview No. 8). Clearing
House Unterricht - TUM School of Education (2014). https://www.clearinghouse.edu.tum.
de/wp-content/uploads/2017/07/CHU_KR_8_Karich_2014_Lernersteuerung-Software.pdf.
Accessed 17 May 2019
11. Kerres, M.: Mediendidaktik. Konzeption und Entwicklung digitaler Lernangebote, 5th edn.
De Gruyter, Oldenburg (2018)
Reflection of HCI in Foreign Language Teaching
Sarka Hubackova(B)
1 Introduction
The biggest focus of Human-computer interaction (HCI) as a multidomain discipline
is put on a user interface. The interconnection with IT science world shall be obvious.
We even today in times of modern technologies, even better computers and cell phones
know, or at least suspect, that there is always something remaining for improving. We
aim to mediate the communication between human and computer in such a way, that it
would be as most intuitive and natural.
The very beginning of HCI is often meant 1857 with the first mention on ergonomics.
With the development of industrialization, scientists examined unfriendly working con-
ditions of workers operating the machines. At the beginning of the 20th century first
studies appeared focused on safety and health protection during work with various types
of machines.
The first document to be considered a direct ancestor of today’s HCI comes from
Russia from 1930. It is dealing with human factor analysis within the airplane cockpit.
In the fifties of the 20th century, human factors and ergonomics became key domains
worldwide. With growing automatization of working processes, a new milestone for
HCI occurred in the sixties – standardization and international certification, which are
until today considered an important component of industrial production and quality
management. Until the second half of the seventies, the only people to come in contact
with computers were IT professionals or enthusiasts. This was totally changed with the
development of personal computers followed by personal software (as productive appli-
cations, interactive games) and platforms (hardware, programming languages, operation
systems). Extended opportunities made a potential user of IT of almost anyone.
HCI is a very broad and multi-disciplinary domain. It is often very tightly bounded
to another domain, where it solves different domain-specific tasks. In informatics, for
instance, it may deal with application or user interface design, while in connection with
psychology it analyses user behavior.
Today we already cannot imagine life without media, the Internet and especially
mobile phones and applications. All these technologies bring us a brief information
overview, facilitated data processing, and last but not least fun. Multimedia technolo-
gies have become an integral part of the present time and through their properties have
begun to influence and change the whole society. Information technology and Inter-
net management allow us to overcome distances in communication, transfer and share
information incredibly fast, faster than any other available technology. They give us new
opportunities and their rapid development opens up new ways constantly.
Multimedia supported teaching is nowadays connected mainly with a computer.
Teaching materials are created with a computer and presented also.
Students consider ICT as a commonplace, as an integral part of their lives. The
organisation of the study is influenced by this fact, for example, consultations are agreed
through email correspondence, submitting of assignments also works on an online basis.
The creation of a multimedia application puts higher demands on the creator, then
common application. Involving different types of data means to contemporary master
several professions at once. Whether it is a presentation or tutorial, it is always necessary
to prepare quality drawings or pictures. So the creator becomes an artist or photographer.
Good sound of the program requires knowledge of a sound man, whether technical or
music, not to mention the difficulty of creating quality animation [1].
The interactive multimedia is a phrase, that defines a new wave of computer software,
that primarily deals with information provision. The multimedia component is charac-
terized by the presence of text, image, audio, animation, and video, that are organized
54 S. Hubackova
into an intelligible program. The part “interactive” relates to the process of empowering
the user to control the environment with a computer [2].
2 Methods
To identify the relationship of students to ICT and to determine students’ views on
teaching supported by ICT, we used the method of a questionnaire. All questions were
closed and offered a choice of several options. The research was initiated in the year
2017/18 and had two phases.
3 Findings
The process of remembering is dependent on external and internal influences and also
on how the current information is being received and processed. How much people
can remember is shown by Dale’s cone of learning [3, 4]. On the model we can see,
that by a simple reading of study (or other) material one can remember only 10% of
information. With the cooperation of sight and hearing the successfulness of learn-
ing increases substantially. This all highlights the advantages of multimedia supported
teaching (Fig. 1).
Learning with multimedia support has gradually become a global trend. Smartphones
and tablets open up almost limitless possibilities for teachers. Educational applications
may serve as didactical tools to explanation or practice of curriculum.
Reflection of HCI in Foreign Language Teaching 55
– Simplicity
– Accessibility - varied content is available in many ways.
– Immediacy and ubiquity - the student is able to find desired information wherever.
– Cooperation - mobile devices simplify group education.
– Interactivity - contact, and cooperation of professionals, teachers or colleagues in
synchronous or asynchronous communication.
– Situation - education can be integrated into everyday life, to places where we face
authentic problems.
– Context - is produced by students through interaction with the surrounding world.
– Convenience - m-learning is the paperless, transportable and interactive educational
environment.
In our lessons, we use multimedia and multimedia features quite commonly. Mul-
timedia online courses we create ourselves, form a part of German-language teaching
already many years. These courses usually serve as a supplement to face-to-face teaching
and are therefore blended learning. In our courses, we use already prepared multimedia
materials we draw from original German-language sources - eg. Deutsche Welle [5, 6].
The most proving show the audio texts, which offer besides audio texts also text files
and listening exercises. Very suitable are also video files, which are supplemented by
texts and exercises created by us. To the courses, we include also many of the students
PPT presentations [7, 8].
56 S. Hubackova
We tried to explore the current state of the use of multimedia in teaching and
learning with research. We focused primarily on what devices students use most fre-
quently. It turned out, that while on secondary schools tablets are the most used, at
university students use laptops in combination with smartphones [9, 10]. This can be
probably explained by the intensity of study when tablets are no longer sufficient and
simultaneously the availability and constant presence of smartphones is well usable
(Figs. 2 and 3).
Dictaphone 1%
TV 4%
DVD/video 4%
Textbook 15%
Smartphone 18%
Internet 27%
Multimedia courses 31% (Fig. 4).
The next question was: More suitable for you is contact learning or teaching sup-
ported by the multimedia courses, that is blended learning? The method of blended
learning pivoted in both groups in a single-minded way.
Reflection of HCI in Foreign Language Teaching 57
4 Discussion
Our survey has shown that modern methods are popular both with full-time students and
distance students and that the entire educational process depends mainly on the teacher’s
approach to teaching and his usage of methods in its implementation.
eLearning will probably never entirely replace the face-to-face form of education,
especially in areas, where personal contact between teacher and student is essential.
In spite of this fact, it became a very important part of the education process. There
are particularly various ways of using eLearning in the scope of further education.
Contemporary, mobile technologies are on a huge rise. They are faster, technologically
more sophisticated and they deal with almost the same abilities and tools like notebooks
and PCs. But they are smaller, therefore more utilizable on the go.
5 Conclusion
The multimedia courses are a very good motivation tool. They surely support all educa-
tion, foreign language teaching inclusive. The blended learning method is very popular
among the students. We constantly complete the current materials from the foreign web
pages. We have prepared the language multimedia courses for beginners or advanced
students of German. We can offer the courses of ordinary, business, and banking lan-
guages. Based on our praxis we consider the form of combined education as a very
suitable one and we count on its further extension to other fields.
Reflection of HCI in Foreign Language Teaching 59
Acknowledgment. This study is supported by the IGS project 2019, run at the Faculty of
Informatics and Management, University of Hradec Kralove, Czech Republic.
References
1. Sokolowsky, P., Šedivá, Z.: Multimédia: Současnost budoucnosti. Grada, Praha (1994).
204 s.
2. Sanjaya, M., Ramesh, S.C.: Interactive Multimedia in Education and Training. Group
Publishing, Hershey (2005). 421 s.
3. Anderson, H.M.: Edgar Dales Cone of Experience (2016)
4. Dales Cone of Experience. http://www.mywallpaper.top/edgar-dales-cone-of-learning.html
5. Čáp, J.: Psychologie pro učitele, Praha (2001)
6. Frydrychova Klimova, B.: Blended learning, in Research, Reflections and Innovations in
Integrating ICT in Education, Lisboa (2009)
7. Pikhart, M.: New horizons of intercultural communication: applied linguistics approach.
Procedia Soc. Behav. Sci. 152, 954–957 (2014)
8. Pikhart, M.: Multilingual and intercultural competence for ICT: accessing and assessing
electronic information in the global world. In: Choroś, K., Kopel, M., Kukla, E., Siemiński,
A. (eds.) MISSI 2018. AISC, vol. 833, pp. 273–278. Springer, Cham (2019). https://doi.org/
10.1007/978-3-319-98678-4_28
9. Pikhart, M.: Communication based models of information transfer in modern management
– the use of mobile technologies in company communication. In: Innovation Management
and Education Excellence through Vision 2020, IBIMA 2018, pp. 447–450 (2018)
10. Hubackova, S.: Foreign language teaching with WebCT support. Procedia Soc. Behav. Sci.
3(2010), 112–115 (2010)
11. Khan, B.H., Granato, L.A.: Program Evaluation in eLearning. http://asianvu.com/
digitallibrary/elearning/elearning_program_evaluation_by_khan_and_Granato.pdf.
Accessed 23 Sept 2017
Expert-Oriented Digitalization of University
Processes
Abstract. Digitalization challenges the way business processes are seen. The
potential for enhancement is recognized even in business areas that traditionally
have little to do with IT. Even though universities have long traditions of how work
is organized, they have not been eager to adopt digitalized processes. Because core
processes of universities rely on highly skilled experts, digitalizing processes is not
as straightforward as in more mechanical work. We developed an expert-oriented
digitalization model (EXOD) for university processes’ digitalization and tested it
using a case study. After digitalizing a core process, we interviewed the experts
involved. The results show the usefulness and adaptability of the model. Based on
the results, we recommend future studies be done to refine and test the model more
comprehensively. Also, based on the adaptability of the model, we recommended
it as a baseline for university process digitalization projects in general.
1 Introduction
Due to digitalization, the importance of information systems (IS) has grown in business
areas that are not normally considered to be IT-oriented [1]. Universities are no exception,
even though some university processes have a long and rather changeless tradition,
inherited from as far back as the 15th century. Long traditions could be seen as an
obstacle for digitalizing university processes, but there are also other obstacles. The
core education processes of universities rely heavily on expert work; the amount of
mechanical work is rather small. Experts with strong opinions and expertise combined
with high autonomy have to be taken into account in university digitalization projects.
In this study, we selected one of the core processes of every university: the thesis
process. Even though the thesis process is critical for universities, it is not usually
considered as a systematic process, but more as the repetition of unique handicraft done
with the supervisors’ best skills and will.
The challenges of the thesis process have been recognized, and some related work has
been done in the areas of both quality improvement and ICT system support [2–4]. One
of the tested thesis process support systems is SciPro [5], which has been studied from
the viewpoints of the student and supervisor interaction and the effective implementation
of the process [3, 6]. Scaling the process for a larger scale implementation has also been
studied from a quality [7] and resource management viewpoint [5].
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 60–69, 2020.
https://doi.org/10.1007/978-3-030-38778-5_8
Expert-Oriented Digitalization of University Processes 61
However, in addition to the quality and resource aspects, the issues in scaling the pro-
cess include integrations with other (core) processes and both manual and ICT systems.
Existing work identifies the thesis process as a core activity in universities [5], but the pro-
cess and systems integration at the organizational level has not been discussed in detail.
These have a considerable impact on, for example, the level of automation as well as infor-
mation availability and quality of the organizational level. In addition, it seems that prior
literature considers the thesis process only as a research process [see, e.g., 3], which is
not the reality in all universities. Other types of theses are also used [see, e.g., 4], and
in different disciplines, different aspects are emphasized within the same type of thesis.
Furthermore, if Davenport’s [8] knowledge work classification is applied, thesis super-
vising can be classified as an expert model of knowledge work, where experts organize
their work individually and are not ready to consent to a mechanical, “cookbook” app-
roach [8]. The thesis process and the supporting IS have to be flexible enough to allow
efficient supervision of different types of theses.
Since prior literature considers the process–system integration on a limited orga-
nizational and individual level, we developed an expert-oriented digitalization model
(EXOD) for digitalizing universities’ learning supporting processes.
To test the developed model, we formulated the following research questions:
2 Theoretical Background
2.1 Business Process Development
If digitalization is done just by automating processes as they are, the existing prob-
lems are fixed with IS, and the potential of IS is not exploited. In addition, as Argyris
[9] points out, people seldomly do exactly what they claim to do, and automating the
assumed process brings out this discrepancy: the new IS may follow the known process
model exactly but is not suitable for use [10]. Thus, automating the processes could be
one part of functional stupidity [11], but as Venkatraman [12] points out, in some cases,
it could be the rational choice of an organization to avoid radical changes in processes.
An organization may choose to automatize existing practices only, instead of attempt-
ing big re-engineering projects. According to Venkatraman [12], IT-enabled business
transformation can be classified roughly into two categories: evolutionary levels and
revolutionary levels, where the former needs minimal changes to business processes,
and the latter requires fundamental changes to existing processes. Venkatraman claimed
that with the revolutionary approach, organizations could benefit more, but the costs
(efforts) of the change would be higher as well. So, there is no right or wrong or optimal
level of business transformation; the cost and potential benefits, as well as the enablers
and inhibitors of the organization, should be taken into account, and each case should
be discussed separately [12].
62 R. Kauppinen et al.
Davenport and Short [13] present a five-step model for process redesign. In their
model, the first step is to develop a business vision and process objectives. This is a rather
general level step and should be done as a part of strategic planning. The second step of the
Davenport and Short [13] model is to select a suitable process(es) to be redesigned. They
point out that it is not necessary to go through all processes of organization exhaustively;
it is enough to identify the most important or most problematic process to be developed.
The third step is understand and measure the selected process(es) to find out current prob-
lems and set a baseline for improvements. The fourth step of the Davenport and Short [13]
model is to identify IT levers, i.e. how IS can enhance the current process or enable totally
new kinds of approaches. The last step is to design and build a prototype of the process
by implementing the new process on a pilot basis and modifying as necessary [13].
In process development, it is not enough to consider the organizational level: indi-
vidual levels have to be taken into account, especially in knowledge work [8]. When
Taylor composed his principles of scientific management, the assumptions about humans
were rather mechanistic: replaceable components doing simple, repeatable tasks, and by
optimizing the tasks, the maximum efficiency is achieved [14]. This, however, is shown
to be an oversimplification, and later process-development models, such as business
process management (BPM), emphasize the role of people and culture [15]. People are
more complex than just parts of a machine; they are not fully rational [11, 16], nor are
they are always reliable. According to Argyris [9], there is a difference between what
people say they do and what they really do. This kind of cover-up culture, or inhibiting
loops of organizational learning, as Argyris [9] calls them, hides the real causes of the
problems [9].
As Davenport [8] points out, knowledge work is difficult to structure and seldom seen
as a process. In addition, knowledge workers easily resist instructions and models given
outside and view a formal process approach as a bureaucratic, procedural annoyance [8].
Even though knowledge workers resist change, there are examples in which knowledge
work is significantly improved through process management [8].
Davenport [8] formulated a model of four approaches to knowledge work to clarify
different knowledge work situations (see Fig. 1).
As thesis supervising, like many other university processes, is clearly more about
judgement and interpretation done alone, we place thesis supervising in the “expert mod-
el” category. According to Davenport [8], expert work can be improved with processes,
but workers themselves easily resist the change and strict, cookbook-type process mod-
els. So, instead, expert model processes should consist of higher level guidelines, giving
expert workers enough flexibility to decide how to do the actual work [8]. To overcome
the expert workers’ resistance and to structure their work, Davenport [8] recommends
finding a way to embed a computer in the middle of the work process. However, IS
should not be an obstacle for experts to reach their full potential [17].
should be considered. A rather famous example of the latter belief is Kotter’s eight-step
model [25], in which the change sticks only when “new behaviors are rooted in social
norms and shared values” [25]. In Kotter’s eight-step model, the idea is that the change
process goes through change steps, and skipping steps creates problems. The steps of the
model are 1. establishing a sense of urgency, 2. forming a powerful guiding coalition,
3. creating a vision, 4. communicating the vision, 5. empowering others to act on the
vision, 6. planning for and creating short-term wins, 7. consolidating improvements and
producing still more change, and 8. institutionalizing new approaches [25].
As process change often manifests itself as a new IS, the success of the new IS
represents the success of the process change. It is, therefore, natural to consider how
new IS are taken in use. There are related theories, for example, technology acceptance
model (TAM), which can be applied as well. According to TAM, the perceived usefulness
and perceived ease of use affect the behavioural intention to use a system [26]. So, to get
users to use a new system, a user has to be made to see that the system increases work
performance and can be used without additional effort.
As mentioned, not all problems in process change are easily seen [11] or recognized
[9]. Thus, it is not possible to deal with them early, and Kotter’s eight-step model is
hard to apply as such. Cooper and Zmud [27] proposed an IT implementation process,
where diffusion of IT does not happen all at once, but as a gradual process. We claim that
Cooper and Zmud model is useful when actual process change (and the supporting IS)
is implemented, whereas Kotter’s model is effective when the change is communicated
to users, keeping the objectives of TAM (usefulness and ease of use) in mind.
Based on the theories of business process and IS development and change management,
we formulated an expert-oriented digitalization model (EXOD) for knowledge work,
especially for university processes. EXOD has four main steps:
3 Methodology
3.1 Research Method
In the case study research, we followed the recommendations of Yin [28]. We used
four data collection sources extensively that Yin [28] recommends, namely documenta-
tion, archival records, participant-observation, and interviews. In the analysis, the main
emphasis was on the interviews; the other sources were considered complementary.
Since one of the researchers was responsible for the thesis process development and
another for the development of IS (Konto) supporting it, we had access to the thesis
process development, as well as all of Konto’s development documentation (process
models, notes, product backlogs, version history, plans, e-mails, guidelines). We also
utilized Konto’s logs and registers as supporting data to understand the actual usage of
the IS. In addition, as supervisors and thesis coordinators, we also used and guided the
use of the digitalized process and made participant observations during the process.
The interviews were done by applying an interview method protocol developed by
Dahlberg, Hokkanen, and Newman [29]. During an interview, questions were presented
on screen either face-to-face or via a video call to the interviewee. The interviewer
recorded and presented the responses immediately before moving to the next question.
Recording the responses gave interviewees the ability to validate the typed answers
immediately. The interviews can be described as expert interviews [30].
The interview had two parts; 27 participants were interviewed. Nearly half, 13
responded to the first and, almost all, 25 to the second part. The interviewees were
chosen based on their above average activity around Konto. In this study, data from the
first part, covering the process of digitalization, is analyzed. The second part, focusing
on the resulting process and tool will be analyzed in our future work.
Of the 13 interviewees, 10 performed a single role, two performed two roles, and
one performed three roles. The fields of expertise covered administration (4), degree
program management (2), thesis coordination (5), and thesis supervision (6).
The digitalization part of the interview consisted of identifying the role of the inter-
viewee and responding to six open-ended questions, as well as an opportunity to provide
open comments. The answers were coded based on the theory presented in Sect. 2 (pro-
cess development, IS development, and change management) and on RQ1 and RQ2
(expectations, experiences, and realization of involvement). One code (service promise)
emerged based on the answers.
A thesis process as a core activity [5] is often considered relatively simple: the supervisor
as the expert advices, and the student writes the thesis [3]. In practice, the process is more
complicated [2, 6]. For example, in our case study, the process at the Haaga-Helia Uni-
versity of Applied Sciences (HH) included other experts, such as the thesis coordinator
(organizes information sessions, checks students’ thesis ideas, and assigns supervisors),
degree program management (oversees supervisors’ and coordinators’ workload), and
the administration (publishes the resulting thesis and records the grade).
66 R. Kauppinen et al.
This process was digitalized using the EXOD model. In initiation (before 2014), HH
described its core processes, revealing that the thesis process was the most complicated.
The benefits of digitalization were apparent, so work started with experts on the process
and IS development with a process re-engineering emphasis (2014–2017). The resulting
process has six phases. The main requirements for the IS were integration with data
sources, automatic data transfer and being a modern platform supporting mobile use.
A hybrid approach [19] that supported the expert involvement was selected, and the
Konto tool was developed in the IS development emphasis (2016–2019) based on the
requirements. Changes to the process were implemented and communicated. After the
fall 2018 test period, the Konto tool was launched for full use with thesis projects starting
from January 2019, resulting in stabilization (from 2019) where the digitalized process
is being committed as a normal activity. Refinements are done as needed.
4 Results
Regarding the experiences (RQ1), the majority of interviewees (10 out of 13) had formed
expectations early, after being involved in the digitalization. From administrative, man-
agement, and coordination viewpoints, as expressed in interviews (translated to English),
process visibility (on every level of the organization), process automation (automating
parts of the process), and statistics recording (getting rid of manually keeping track of
supervisors and their resources) were considered especially important. The supervisors
and coordinators emphasized the change in communication and the usefulness of the
single platform (fewer e-mails when the communication and materials are in the same
place), and the transparency (the supervision is visible).
Of the interviewees, four out of 13 identified only positive experiences, eight identi-
fied both positive and negative, and one identified only negative. The experiences were
higher in number and more detailed for coordination and supervision, while experiences
of the administration and management were fewer and more general. The positive expe-
riences were related to the model (extremely useful, agile model that utilizes in-house
competencies well and is generalizable to similar, well-scoped development efforts), the
involvement (it has been valuable, being able to participate and try out, which also helps
in commitment to the result), and influencing the result (the needs of the users have been
taken into account).
The negative experiences were doubts about the coverage of the involvement (the
piloting phase could have been longer, and more people could have been involved), and
coping with incompleteness (some may have felt insecure due to the changes). It is worth
noting that the interviewee stating only negative experiences still felt the participation
itself positive and considered related work on the service promise to be helpful. Another
interviewee also mentioned the service promise as supporting the involvement.
The realization of involvement (RQ2) was difficult to pinpoint. While eight out of
13 interviewees acknowledged having development ideas, most comments were general
and did not name concrete examples. Instead, they were showing trust in taking the
ideas into account (there may be something that I also have pointed out, but it is hard to
specify a single one). Only a few could name a concrete and implemented idea (it was
not possible to send a message in a certain situation, but now it is).
Expert-Oriented Digitalization of University Processes 67
The interview data, Konto logs, documentation, and observations all confirmed that
the EXOD model performed well in digitalizing the expert-driven thesis process. The
findings met the goals set for the EXOD model: the experts felt that they had been listened
[8, 25], the developed IS decreased the workload of experts [8], it was easy to use [26],
and it ensured that the process was followed [8]. The level of process development was
meaningful [12, 13] and gave a good basis for the development of the IS. The selected
IS development method [23], the hybrid approach [19], was suitable in this case.
The best results were achieved in change management. Thesis supervision is consid-
ered to be personal expert work [8], and external interventions, such as process enforcing
and automation, are often considered undesirable. However, in this case, the experts felt
they could affect the outcome [25] and automation, for example, was seen useful as it
reduced mechanical work and clarified information handling [8, 13, 25, 26].
However, there is still room for improvement. Some experts perceive continuously
changing IS to be confusing, so, plan-driven development could be emphasized. In
addition, some felt that pilots were short, so feedback could be collected over a longer
period and from a larger user group. Also, in some cases, users claim to follow the
process but implementing the IS revealed that they actually do not [9]. While this cannot
be fully avoided, it should be considered in EXOD model steps 2 and 3 by engaging
experts to pursue objectives. In addition, some supervisors emphasize their expertise in
supervising, while others see it more as routine work, so compromises need to be made.
Interestingly, the EXOD model seems to produce committed change agents [24],
even though it was not an explicitly pursued objective. This effect should be studied and
further developed to make it more robust. Overall, we recommend future studies be done
to refine and test the EXOD model more comprehensively. Also, based on the suitability
for processes with high actor expertise and autonomy, we recommended it be used as a
baseline for university process digitalization projects in general.
68 R. Kauppinen et al.
References
1. Borg, M., Olsson, T., Franke, U., Assar, S.: Digitalization of swedish government agencies —
a perspective through the lens of a software development census. In: International Conference
on Software Engineering (2018)
2. Aghaee, N.: Finding potential problems in the thesis process in higher education: analysis of
e-mails to develop a support system. Educ. Inf. Technol. 20, 21–36 (2015). https://doi.org/
10.1007/s10639-013-9262-z
3. Karunaratne, T.: Blended supervision for thesis projects in higher education: a case study.
Electron J. e-Learn. 16, 79–90 (2018)
4. Lagstedt, A.: Diary thesis as a tool for professional growth and for co-operation between
universities and business. In: Proceedings of the 2015 UIIN Conference, June 2015, Berlin,
Germany (2015)
5. Hansson, H.: How to produce quality theses at universities in a large scale: SciPro IT system
- Supporting the Scientific Process. In: Proceedings - Frontiers in Education Conference, FIE
(2014)
6. Hansen, P., Hansson, H.: Optimizing student and supervisor interaction during the SciPro
thesis process – concepts and design. In: Li, F.W.B., Klamma, R., Laanpere, M., Zhang, J.,
Manjón, B.F., Lau, R.W.H. (eds.) ICWL 2015. LNCS, vol. 9412, pp. 245–250. Springer,
Cham (2015). https://doi.org/10.1007/978-3-319-25515-6_23
7. Larsson, K., Hansson, H.: The challenge for supervision: mass individualisation of the the-
sis writing process with less recourses. In: Online Educa Berlin 2011-17th International
Conference on Technology Supported Learning & Training, pp. 2007–2009 (2011)
8. Davenport, T.H.: Process management for knowledge work. In: vom Brocke, J., Rosemann,
M. (eds.) Handbook on Business Process Management 1. IHIS, 2nd edn, pp. 17–35. Springer,
Heidelberg (2015). https://doi.org/10.1007/978-3-642-45100-3_2
9. Argyris, C.: Organizational learning and management information systems. Acc. Organ. Soc.
2, 113–123 (1977)
10. Lagstedt, A., Dahlberg, T.: Requirements engineering as a part of business process and infor-
mation system development. In: Proceedings of the Seventh International Conference on
Well-Being in the Information Society: Fighting Inequalities (WIS 2018), pp. 35–39 (2018)
11. Alvesson, M., Spicer, A.: A stupidity-based theory of organizations. J. Manag. Stud. 49,
1194–1220 (2012). https://doi.org/10.1111/j.1467-6486.2012.01072.x
12. Venkatraman, N.: IT-enabled business transformation: from automation to business scope
redefinition. Sloan Manage. Rev. 35, 73–87 (1994)
13. Davenport, T.H., Short, J.E.: The new industrial engineering: information technology and
business process redesign. Sloan Manage. Rev. 31, 11–27 (1990)
14. Taylor, F.W.: The Principles of Scientific Management. Harper & Brothers Publishers, New
York (1913)
15. Brocke, J., Sinnl, T.: Culture in business process management: a literature review. Bus.
Process. Manag. J. 17, 357–377 (2011). https://doi.org/10.1108/14637151111122383
16. Simon, H.A.: Administrative Behavior, 4th edn. The Free Press, New York (1997)
17. Wenger, E.C., Snyder, W.M.: Communities of practice: the organizational frontier. Harv. Bus.
Rev. 78(1), 139–145 (2000)
18. Moe, N.B., Aurum, A., Dybå, T.: Challenges of shared decision-making: a multiple case study
of agile software development. Inf. Softw. Technol. 54, 853–865 (2012). https://doi.org/10.
1016/j.infsof.2011.11.006
19. Theocharis, G., Kuhrmann, M., Münch, J., Diebold, P.: Is Water-Scrum-Fall reality? On the
use of agile and traditional development practices. In: Abrahamsson, P., Corral, L., Oivo,
M., Russo, B. (eds.) PROFES 2015. LNCS, vol. 9459, pp. 149–166. Springer, Cham (2015).
https://doi.org/10.1007/978-3-319-26844-6_11
Expert-Oriented Digitalization of University Processes 69
20. Hansen, S., Lyytinen, K.: Challenges in contemporary requirements practice. In: Proceedings
of the Annual Hawaii International Conference on System Sciences, pp. 1–11 (2010). https://
doi.org/10.1109/HICSS.2010.98
21. Hastie, S., Wojewoda, S.: Standish group 2015 chaos report - Q&A with Jennifer Lynch. In:
InfoQ, blog (2015). https://www.infoq.com/articles/standish-chaos-2015. Accessed 22 Feb
2018
22. Dahlberg, T., Lagstedt, A.: There is still no “ Fit for All” IS development method : business
development context and IS development characteristics need to match. In: Proceedings of
the 51st Hawaii International Conference on System Sciences (2018)
23. Lagstedt, A., Dahlberg, T.: A contingency theory motivated framework to select information
system development methods. In: Pacific Asia Conference on Information Systems, pp. 1–14
(2018)
24. Cameron, E., Green, M.: Making Sense of Change Management, 2nd edn. Kogan Page Ltd.,
London (2009)
25. Kotter, J.P.: Leading change: why transformation efforts fail. Harv. Bus. Rev., 59–67 (1995)
26. Davis, F.D., Bagozzi, R.P., Warshaw, P.R.: User acceptance of computer technology: a com-
parison of two theoretical models. Manage. Sci. 35, 982–1003 (1989). https://doi.org/10.
1287/mnsc.35.8.982
27. Cooper, R.B., Zmud, R.W.: Information technology implementation research: a technological
diffusion approach. Manage. Sci. 36, 123–139 (1990). https://doi.org/10.1287/mnsc.36.2.123
28. Yin, R.K.: Case Study Research: Design and Methods, 4th edn. Sage Publications, Thousand
Oaks (2009)
29. Dahlberg, T., Hokkanen, P., Newman, M.: How business strategy and changes to business
strategy impact the role and the tasks of CIOs: an evolutionary model. In: Proceedings of the
Annual Hawaii International Conference on System Sciences, pp. 4910–4919 (2016)
30. Bogner, A., Littig, B., Menz, W.: Introduction: expert interviews – an introduction to a new
methodological debate. In: Bogner, A., Littig, B., Menz, W. (eds.) Interviewing Experts,
pp. 17–42. Palgrave Macmillan, New York (2009)
Itinerant Virtual Museum: An Innovate
Technique to Learn Ancient History
1 Introduction
The ancient civilizations of the world (Sumerian, Egyptian, Greek and Roman), con-
stitute one of the most complex topics of the Costa Rican Social Studies curriculum.
The history contents related to these civilizations are relevant because of their legacy to
humanity, in fields like the cultural, architectural, social, political, economic and geo-
graphical. However, the learning such contents and history in general present a series of
difficulties in some places in the world such as Costa Rica.
The Social Studies subject is taught trough a behavioristic lecture and does not allow
to develop to critical skills or 21st century skills in the students. This causes a negative
perception towards the learning process, and more specifically towards learning about
history. Some additional consequences are memorization [7–16, 19], the episodic narra-
tive, magistral class [6, 12], the lack of historical understanding [7–16], it is anachronistic
[6, 13], it is boring [16, 18], it is useless [16] and teachers’ methodology [15, 16]. The
previous difficulties seek to be solved with different methods by teachers and researchers,
who understand that contextualization to digital natives to learn is the key approach to
understanding learning of history in Social Studies.
In this research an innovate technique is proposed to learn history in a contextualized
style of the current society and allows learning through experiences, this method is the
use of virtual museums with virtual lenses. In this study, this new type of museum with
virtual lenses will be called Itinerant Virtual Museum (IVM). The reason of the name
is due to the museum that can be mobilized (itinerant means that it travels from one
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 70–75, 2020.
https://doi.org/10.1007/978-3-030-38778-5_9
Itinerant Virtual Museum: An Innovate Technique to Learn Ancient History 71
place to another) to any kind of school or place inside or outside of the country, to
learn about specific historical topics. In some Costa Rican museum valijas didácticas
(teaching bags) are used, which have some museum elements to learn history and they
can be taken around the country to some schools or institutions that do not have financial
resources or the time. That is how the idea to create an IVM that can travel to communities
was born, it is not only to use the platform where the museum is, but the use of the lenses
and a device with Wi-Fi in case internet is needed in the community (Fig. 1).
Fig. 1. Workshop 1
The purpose to use Virtual Museum is to benefit education and history teaching,
like the approach to this area knowledge [5], motivational space and attractiveness
[5, 17], promoting cultural legacies and learning experiences [5, 11], developing skills
as creativity, critical thinking, autonomy, digital competences among others [1, 11],
observing events or phenomenon that are difficult to understand [8, 20], saving time and
money [17]. In brief, the IVM could become and innovate tool to learn ancient history
and it could solve some problems when someone is learning history in Social Studies.
2 Methodology
The nature of the research is qualitative and it uses a methodology called participative
design, also known co-design. This methodology seeks to solve a problem in an inno-
vative way with the contribution of the users (in this research they will be students).
This methodology is Scandinavian, and it applies to different areas [9] such as the edu-
cation, where to evaluate an educational need by a prototype is essential to approach
an educational innovation [3] with the cooperation of the users, which is imperative
[2–4, 14]. The co-design phases applied in this research [3] are Preparation (determine
the research problem), Exploration (seek to have greater clarity regarding the research
problem), Vision (users and teachers explain how they could solve the problem), Oper-
ationalization (researcher and users create IVM prototypes) and Evaluation (assessment
of experiences and project is improved and re-evaluated).
72 L. Díaz Granados
The first group is a sample of seven students (three men and four women) from a high
school in Santo Domingo de Heredia. They are thirteen and fourteen years old and
belongs to middle socioeconomic class. It is important to mention the students in the
research will be named users. One of the reasons why the sample is so small, is due to the
administrative permissions in the high school (for the loss classes time in other subjects
and the schedule provided by the principal), the use of virtual lenses (the school only has
four devices) and lack of space. The application to small sample allows a manageable
task when the researcher applies methods like interview or the observations [10]. The
second group is formed for a team of Social Studies teachers who will evaluate the IVM
in the workshop number four, to value the ancient history learning. The third group are
the experts, who will evaluate the technical part of the product in the fields as content,
design and navigation of the IVM. In this group are three professionals like a graphic
designer, an historian and a virtual specialist.
The qualitative techniques to data collect applies in the study are five focus group with
users, teachers and experts (the first one asked for history, difficulties, solutions, tools and
the next ones ask for navigate, content, design, experiences, preferences of the IVM).
Four participant observations and four participative workshops to users, teachers and
experts. These workshops test the IVM prototypes and each Workshop has methods
to collect data (Thinking loud or do small assignment in the IVM). Data has been
transcripted, classified, reviewed and compared several times to obtain a reliable result.
There are two important moments missing to finish the study, these moments are the last
test of prototype with users (third workshop) and the evaluation from the experts (fourth
workshop).
2.3 Results
The results obtained so far were achieved implementing the last two co-design phases,
however, we are presenting results in advance with techniques and analysis of data collect
to this point.
1. Alternative tools to learning history: The users-students in the first focus group and
the second workshop applied, mentioned that teachers must use different method-
ologies and tools to teach history in Social Studies. The seven students said that
they like the subject, but six of them get bored in class and when they study for the
exam, they consider the contents or topics very broad and boring. All the students
agree that is necessary to apply new tools in class. The examples provided for the
students after a web research of strategies or tools to help the learning history are:
videos, civilizations o culture play, virtual walks, google earth, 360 grades pictures
and Spanish information, because most the information found is in English and the
information in Spanish is small and is not interesting.
Itinerant Virtual Museum: An Innovate Technique to Learn Ancient History 73
2. Use of visual information: The focus group in the workshop one and two, provide
information about how the students and users don’t like the textual information,
let alone if the information is vast. In both workshops applied, the students look for
information in small videos, summary and specific images, places and information
with a lot of colors and the information in videos or images is inside of the IVM,
because in the first prototype some information was in a website outside of the IVM;
they preferred to stay only in the IVM and could access all the information there.
3. Immersion research: In the first focus group and second workshop, four of six users
indicate the preference to move inside the IVM, the words used by two of them were
“Poder caminar” (be able to walk). Moreover, the users would like play inside of
IVM something related with the civilization when they selected some element in the
museum, and they would like to visit places of each ancient civilization.
4. More guidance: In the first workshop, for many of the users it was complicated to
use the virtual lenses, the pointer and to search information in the IVM. Four users
of six consider it complex the use of the pointer and one user suggest that a guide
was necessary to learn how to learn used it. Even though in the second workshop,
the use of the pointers and virtual lenses were more natural and simpler, one user
still said it was difficult to navigate inside of the IVM and to use the pointer. Most
of the users consider important to have a guide inside of the museum, in physical
appearance, audio or information.
5. Viable as a learning tool: Mostly of the users consider technology as a tool to help
in the learning process like the IVM can support the history learning. Because the
lesson and the learning became a dynamic, interactive and interesting space and
process, this motivation allows to learn history, because is not boring. In the third
focus group, one question was about learn history mediated by IVM, and one of the
users said “Entonces al ser menos aburrida no nos va a costar tanto aprender” (So,
being less boring, it is not going to be difficult to learn).
3 Conclusion
The research is in the final phases and the information presented is the first clue to
understand how the application of new learning tools can approach positive results.
The use of IVM can incite motivation to learn ancient history or history in general and
change the negative perception of some students about Social Studies. This learning tool
can work the difficulties to understand learning about history, because there is a closer
approach to history, development of empathy, interest, pleasure, contextualization to
actual society and innovation. This research is expected to became in a learning tool for
ancient history or others subjects. The research done so far is a platform with information
of four ancient civilization so far away from Costa Rica (one of the reasons to explain
the difficulty to learn this topic) with the most user’s suggestions, opinions, preferences
and elements that they like or dislike. Although the study has been satisfactory, some
improvements must still be made, apply the last two workshops and analyze the data
obtained again, because is the only way to support the idea that IVM is a learning tool
to understand ancient history.
74 L. Díaz Granados
4 Limitations
Among some limitations so far in this research after to apply the mostly phases are: Few
technological devices such as lens (only for), the high school’s internet has blocked a
lot of websites, there is no a stimulating place to apply the IVM prototypes with the
users, problems continue with the use of the pointer, limitations in the platform used to
create the IVM (sound, levels of navigation, small areas to visualize information, cannot
export); difficulties to find places to take 360° pictures to IVM, and difficulties to find
and manuals to add in the IVM, because this information is in other languages.
References
1. Ahmed Ismaeel, D., Al-Abdullatif, A.: The impact of an interactive virtual museum on stu-
dents’ attitudes toward cultural heritage education in the region of Al Hassa, Saudi Arabia.
iJET 11, 32–39 (2016). https://doi.org/10.3991/ijet.v11i04.5300
2. BakirlioğLu, Y., OğUr, D., DoğAn, C., Turhan, S.: An initial model for generative design
research: bringing together generative focus group (GFG) and experience reflection modelling
(ERM). Des. Technol. Educ. Int. J. 21(1), 40–50 (2016)
3. Barbera, E., García, I., Fuertes-Alpiste, M.: A co-design process microanalysis: stages and
facilitators of an inquiry-based and technology-enhanced learning scenario. Int. Rev. Res.
Open Distrib. Learn. 18(6) (2017). https://doi.org/10.19173/irrodl.v18i6.2805
4. Burkett, I.: Co-designing for social good: the role of citizens in designing and deliv-
ering social services. In: Burkett, I. (ed.) An Introduction to Co-Desing, pp. 3–
11. Knode. https://www.yacwa.org.au/wp-content/uploads/2016/09/An-Introduction-to-Co-
Design-by-Ingrid-Burkett.pdf. Accessed 2018
5. Elisondo, R., Melgar, M.: Museos y la Internet: contextos para la innovación. Innovación
educativa 15(68), 17–32 (2015)
6. Folgueira Lombardeo, P.: Dualidad de la Historia en el mundo 2.0. Tiempo y sociedad 18,
133–143 (2015)
7. Gloria Zúñiga, C.: ¿Cómo se ha enseñado historia en Chile? Análisis de programas de estudio
para enseñanza secundaria. Pensamiento educativo. Revista de investigación educacional
latinoamericana, 119–135 (2015). https://doi.org/10.7764/pel.52.1.2015.9
8. Gonzalez Yebra, O., Aguilar, M., Aguilar, F., Lucas, M.: Co-design of a 3D virtual campus
for synchronous distance teaching based on student satisfaction: experience at the University
of Almería (Spain). Educ. Sci., 1–16 (2019). https://doi.org/10.3390/educsci9010021
9. Gros, B., Zhang-Yu, C., Ayuste, A., Escofet, A.: La apropiación de los dispositivos móviles
en ciudadanos inmigrantes: el segundo nivel de división digital. Athenea Digit. 18(3), e2175
(2018). https://doi.org/10.5565/rev/athenea.2175
10. Hernández Sampieri, R., Fernández Collado, C., Baptista Lucio, M.: Metodología de la
Investigación, vol. 6. McGraw-Hill, México (2014)
Itinerant Virtual Museum: An Innovate Technique to Learn Ancient History 75
11. Islek, D., Asiksoy, G.: The studies conducted regarding virtual museum area: a content analysis
research. World J. Educ. Technol. Current Issues 11(1), 087–093 (2019). https://doi.org/10.
18844/wjet.v11i1.4012
12. Ledezma, I.: Aprendizaje significativo en la enseñanza de la historia. Sede Rodrigo Facio,
Tesis de licenciatura en la enseñanza de los estudios sociales y la cívica de la Universidad de
Costa Rica (2007)
13. Lombardi, A.: La enseñanza de la historia: consideraciones generales. Teoría y didáctica de
las ciencias sociales, 9–23 (2000). ISSN 1316-9505. http://www.redalyc.org/articulo.oa?id=
652/65200502. Accessed 2018
14. Maina, M., Pérez Mateo, M., Guardia, L., Sangrà, A.: Diseño de un curso de formación
docente como práctica educativa abierta. Revista CIDUI, 1–15 (2015). https://doi.org/10.
1016/j.sbspro.2015.07.021
15. Pleguezuelos Castillo, E.: Problemas para aprender ciencias sociales. Tesis de Master Profeso-
rado de Educación Secundaria y Bachillerato, Formación Profesional y Enseñanza de Idiomas
en la Universidad de Almerı̄a (2012). http://repositorio.ual.es/bitstream/handle/10835/2816/
Trabajo.pdf?sequence=1. Accessed 2018
16. Prats, J.: Retos y dificultades para la enseñanza de la historia. In: Sanz Camañes, P., Molero
García, J., Rodríguez González, D. (eds.) La historia en el aula Innovación docente y enseñanza
de la historia en la educación secundaria, pp. 15–32. Editorial Milenio, Leiida (2017)
17. Sabbatini, M.: Centros de ciencia y museos científicos virtuales: teoría y practica. Teoría de la
educación: educación y cultura en la sociedad de la información, vol. 4. Obtenido de Centros
de ciencia y museos científicos virtuales: teoría y práctica (2003)
18. Tok, B.: Learning problems in History subject among the Secondary School-Students of
Papum-pare district of Arunachal Pradesh. IRA-Int. J. Educ. Multidiscip. Stud. 2, 133–139
(2016). https://doi.org/10.21013/jems.v5.n2.p9
19. Valverde, J.: Aprendizaje de la Historia y Simulación Educativa. Tejuelo, 83–99 (2010)
20. Yildirim, G., Elban, M., Yildirim, S.: Analysis of use of virtual reality technologies in history
education: a case study. Asian J. Educ. Train. 4(2), 62–69 (2018). https://doi.org/10.20448/
journal.522.2018.42.62.69
Influence of the Type of Organizations on Their
Readiness for Implementing Industry 4.0
University of Hradec Králové, Rokitanského 62, 500 03 Hradec Králové, Czech Republic
{majid.ziaeinafchi,hana.mohelska}@uhk.cz
Abstract. Many studies have been conducted about Industry 4.0 and also about
readiness and maturity models. The available studies are mostly from the techno-
logical point of view. Nonetheless, the organizational culture dimension of Indus-
try 4.0 has some importance and requires more attention. The goal of this paper is
to find out if the type of an organization impacts the innovative culture and subse-
quently the readiness of the organization for implementing industry 4.0. Results
confirm that in fact the innovative organizational culture according to the index of
organizational culture does depend on the type of an organization, and therefore
type of the organization affects the readiness of the organization for implementing
Industry 4.0.
1 Introduction
The essence of the fourth Industrial revolution is considered and known as Industry 4.0,
and manufacturing sector has been using it lately by utilizing cyber-physical systems
(CPS) in order to grasp high levels of automation (Ziaei Nafchi and Mohelská 2018).
The Cyber-Physical System (CPS) is the basis for smart factories and it makes it possible
to interconnect sensors, machines and IT systems within the value chain throughout the
boundaries of the enterprise (Kopp and Basl 2017).
Many studies have been conducted about Industry 4.0 and also about readiness and
maturity models. The available studies are mostly from the technological point of view.
Nonetheless, the organizational culture dimension of Industry 4.0 has some importance
and requires more attention.
There are three types of organizational culture introduced by Wallach (1983), among
them the innovative culture is believed to be as the type of culture that is enhances
the implementation of Industry 4.0. In other words, Organizations with higher level of
innovative culture are more ready and will have a smoother transition when it comes to
implementing Industry 4.0.
Based on this assumption, the main objective of this study is to find out if the
innovative culture, which is believed to be an important pre-condition for implementing
Industry 4.0, differs in organizations of different types.
2 Theoretical Background
2.1 Organizational Culture
Kopp and Basl (2017) state “Industry 4.0 concept can be characterized as a transforma-
tion of production as separate automated factories into fully automated and optimized
manufacturing environments. Production processes are linked vertically and horizontally
within enterprise systems”.
Thanks to Industry 4.0, manufacturing unique products in terms of excellent quality
has become possible and with a price that is matching the price of mass-produced
products (Nowotarski and Paslawski 2017).
Implementing Industry 4.0 and digital transformation concepts in theory is gradually
significant for manufacturing businesses that are performing in such markets that are
dynamic and competitive. Though in practice, there are a few challenges for organizations
while implementing such concepts since Industry 4.0 is considered to be more as a
concept rather than a ready-to-implement solution; Furthermore, the complex nature of
Industry 4.0 triggers delays to the successful implementation of Industry 4.0 systems in
such a way that they combine all organizational features and levels accurately (Issa et al.
2018).
Suitable resources, accomplished and competent employees and well-organized pro-
cesses, that are appropriately flexible and innovative, are believed to be necessities while
implementing the Smart concept (Odważny et al. 2018).
Lak and Rezaeenour (2018) state “Maturity models can be considered as a structured
collection of elements in which certain aspects of the capability maturity in an organi-
zation are described.” Maturity models are normally used as a tool to conceptualize
and measure maturity of an organization or as a process concerning certain target state
(Schumacher et al. 2016).
The following three implementation phases are suggested by Odważny et al. (2018) to
be distinguished within company: Aspiration phase, Maturity phase, and Smart factory.
Basl and Doucek (2018) state that at the “micro” level, the readiness of the organiza-
tion is essentially an evaluation or assessment of the maturity degree of the organization,
thus the readiness models (maturity models) prevail.
Colli et al. (2018) state “The transformation of the manufacturing sector towards
Industry 4.0 is setting the scene for a major industrial change. Currently, the need for
assisting companies in this transformation is covered by a number of maturity models
that assess their digital maturity and provide indications accordingly.”
To accomplish success in an environment that is as uncertain as Industry 4.0, learn-
ing, training, and innovation capability play substantial roles. Organizational training,
learning, and innovations are intensely dependent on the role employees in the organi-
zation and therefore, organizations need to arrange their strategies in accordance with
to what they expect from their employees (Shamim et al. 2017).
Transparency is another important parameter to be considered as it plays an important
role in rationality, decent governance, and better progress (Ziaei Nafchi et al. 2018).
Influence of the Type of Organizations on Their Readiness 79
Table 1. Results of Statistical analyses of dependence of the culture indexes on the type of
company (C-Czech, I-international, S-state organization)
The innovative culture is the highest in international organizations, and then Czech
organizations are slightly lagging behind. Finally the state organizations with a large
distance from the other two types have the least amount of innovative culture, which
completely makes sense.
Figure 1 clearly shows the differences mentioned before between the different types
of organizations, it is evident that the state organizations have the least innovative culture
and the international organizations have the most bureaucratic culture.
Cronbach’s alpha was calculated for Wallach’s questionnaire based on the dimen-
sions they were associated with in order to check the internal consistency of the question-
naire. The internal consistency of the bureaucratic culture and the innovative culture are
considered to be acceptable and the internal consistency of the supportive culture is good.
80 M. Ziaei Nafchi and H. Mohelská
It is possible to argue why the supportive culture gets weaker the larger the orga-
nization gets in a similar fashion; supportive culture is established on cooperation and
it is relationship-oriented with greater levels of trust, clearly it is more challenging to
establish such values with others in an environment where there are more people.
Based on the findings there is sufficient evidence to conclude that the innovative
organizational culture according to the index of organizational culture depends on the
type of an organization. Hence, the null hypothesis H0 is NOT rejected.
Therefore, it can be said that the type of organization has an impact on the readiness
of it for the implementation of industry 4.0.
Acknowledgement. The paper was written with the support of the specific project 2106/2019
grant “Determinants of Cognitive Processes Impacting the Work Performance” granted by the
University of Hradec Králové, Czech Republic.
References
Armstrong, M.: Armstrong’s Handbook of Management and Leadership: A Guide to Managing
for Results, 2nd edn. Kogan Page, London (2006)
Basl, J., Doucek, P.: Metamodel of indexes and maturity models for Industry 4.0 readiness in
enterprises. In: Doucek, P., Chroust, G., Oškrdal, V. (eds.) IDIMT 2018: Strategic Modelling
in Management, Economy and Society - 26th Interdisciplinary Information Management Talks
Kutná Hora, Czech Republic, 5–7 September 2018, pp 33–40 (2018). ISBN 978-3-99062-339-8
Colli, M., Madsen, O., Berger, U., Møller, C., Wæhrens, B.V., Bockholt, M.: Contextualizing the
outcome of a maturity assessment for industry 4.0. In: IFAC Papersonline, 16th IFAC Sympo-
sium on Information Control Problems in Manufacturing INCOM 2018, vol. 51, pp. 1347–1352
(2018). https://doi.org/10.1016/j.ifacol.2018.08.34
Issa, A., Hatiboglu, B., Bildstein, A., Bauernhansl, T.: Industrie 4.0 roadmap: framework for digital
transformation based on the concepts of capability maturity and alignment. In: Procedia CIRP,
51st CIRP Conference on Manufacturing Systems, vol. 72, pp. 973–978 (2018). https://doi.org/
10.1016/j.procir.2018.03.151
Kopp, J., Basl, J.: Study of the readiness of czech companies to the industry 4.0. J. Syst. Integr.
(1804-2724) 8(3), 40 (2017). https://doi.org/10.20470/jsi.v8i2.313
Lak, B., Rezaeenour, J.: Maturity assessment of social customer knowledge management (SCKM)
using fuzzy expert system. J. Bus. Econ. Manag. 19(1), 192–212 (2018). https://doi.org/10.
3846/16111699.2018.1427620
Naqshbandi, M.M., Tabche, I.: The interplay of leadership, absorptive capacity, and organizational
learning culture in open innovation: testing a moderated mediation model. Technol. Forecast.
Soc. Change 133, 156–167 (2018). https://doi.org/10.1016/j.techfore.2018.03.017
Nowotarski, P., Paslawski, J.: Industry 4.0 concept introduction into construction SMEs. IOP Conf.
Ser. Mater. Sci. Eng. 245(5), 1 (2017). https://doi.org/10.1088/1757-899x/245/5/052043
Odważny, F., Szymańska, O., Cyplik, P.: Smart factory: the requirements for implementation of
the industry 4.0 solutions in FMCG environment - case study. Logforum 14(2), 257 (2018).
https://doi.org/10.17270/j.log.2018.253
Okatan, K., Alankuş, O.B.: Effect of organizational culture on internal innovation capacity. J.
Organ. Stud. Innov. 4(3), 18 (2017)
Rahman, M.H., Moonesar, I.A., Hossain, M.M., Islam, M.Z.: Influence of organizational culture
on knowledge transfer: evidence from the Government of Dubai. J. Public Aff. 18(1) (2018).
https://doi.org/10.1002/pa.1696
82 M. Ziaei Nafchi and H. Mohelská
Schumacher, A., Erol, S., Sihn, W.: A maturity model for assessing industry 4.0 readiness and
maturity of manufacturing enterprises. In: Procedia CIRP, The Sixth International Conference
on Changeable, Agile, Reconfigurable and Virtual Production (CARV 2016), vol. 52, pp. 161–
166 (2016). https://doi.org/10.1016/j.procir.2016.07.040
Shamim, S., Shuang, C., Hongnian, Y., Yun, L.: Examining the feasibilities of industry 4.0 for the
hospitality sector with the lens of management practice. Energies (19961073), 10(4), 1 (2017).
https://doi.org/10.3390/en10040499
Verdu-Jover, A.J., Alos-Simo, L., Gomez-Gras, J.: Adaptive culture and product/service innovation
outcomes. Eur. Manag. J. 36, 330–340 (2018). https://doi.org/10.1016/j.emj.2017.07.004
Wallach, E.: Individuals and organization: the cultural match. Training Dev. J. 37, 28–36 (1983)
Ziaei Nafchi, M., Mohelská, H.: Effects of industry 4.0 on the labor markets of Iran and Japan.
Economies 6(3), 39 (2018). https://doi.org/10.3390/economies6030039
Ziaei Nafchi, M., Mohelská, H., Marešová, P., Sokolová, M.: E-governance: digital trancparency
and the model of interaction within Czech municipalities. In: Doucek, P., Chroust, G., Oškrdal,
V. (eds.) IDIMT 2018: Strategic Modelling in Management, Economy and Society - 26th
Interdisciplinary Information Management Talks, Kutná Hora, Czech Republic, 5–7 September
2018, pp. 41–48 (2018). ISBN 978-3-99062-339-8
Ziaei Nafchi, M.: Industry 4.0 and preparing companies for implementing it. Doctoral dissertation,
University of Hradec Kralove, 2019. Hradec Kralove, Czech Republic: Faculty of Informatics
and Management (2019)
Semantic Competency Directory
for Constructive Alignment in Digital Learning
Designs and Systems
Beuth University of Applied Sciences, Luxemburger Str. 10, 13351 Berlin, Germany
{buchem,konert}@beuth-hochschule.de
1 Introduction
The model of Constructive Alignment by Biggs [1] proposes the systemic alignment of
three elements of learning design, i.e. learning outcomes, learning activities and assess-
ment of learning focused on demonstrating the achievement of the learning outcomes.
The model has been widely used to guide impactful learning design both in context of
traditional teaching and learning practices as well as in technology-enhanced learning,
including Open Educational Practices [2] and Massive Open Online Courses [3]. In
competency-based learning approaches, learning outcomes are defined as skills or com-
petencies, e.g. in form of statements which describe what and how well learners will
able to do after completing a course [4]. Defining learning outcomes as competencies
and the constructive alignment approach have led to improving clarity and transparency
in education [5]. However, some of the practical challenges in the application of con-
structive alignment model remain the development of learning activities and assessment
tasks which are aligned to the learning outcomes [3]. Constructive alignment has been
also discussed in relation to digital micro-credentials, e.g. Open Badges, which can be
constructively aligned with learning outcomes and can serve as evidence that a learning
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 83–88, 2020.
https://doi.org/10.1007/978-3-030-38778-5_11
84 I. Buchem and J. Konert
outcome has been achieved [6]. The problem of alignment of competencies in context of
Open Badges has been the ambiguity of competency descriptions and the lack of seman-
tic metadata of competency frameworks [7]. One of the approaches to this problem are
semantic description of competencies and tools which allow for improved constructive
alignment.
In this paper we describe the semantic competency directory and demonstrate its
value for competency alignment in digital learning designs and systems following the
principles of constructive alignment. We present the competency directory as a technol-
ogy part of the semantic alignment of competencies and describe an application example
from the Open Virtual Mobility project to reference a set of skills in Open Badges. The
paper ends with conclusions and recommendations for future work.
The first prototype of the semantic competency directory was developed in the Open
Badge Network project (OBN, Erasmus+, 2014–2017) as a service for the decentral-
ized search and cross-referencing of linked-data descriptions of competencies [7]. The
next iteration of the competency directory is goal of the Open Virtual Mobility project
(OpenVM, Erasmus+, 2017–2020). Open Virtual Mobility is a strategic partnership for
innovation and the exchange of good practices and aims at enhancing the uptake of vir-
tual mobility in higher education by improving virtual mobility skills. To achieve this,
the Open Virtual Mobility Learning Hub was developed as a learning environment for
achievement, assessment and recognition of virtual mobility skills [8]. The OpenVM
Learning Hub hosts altogether eight different miniMOOCs each dedicated to a specific
virtual mobility skill area. The eight skill areas were identified through a Group Con-
cept Mapping study [9]. The research results were used to define the skill set with eight
main skill areas and respective sub-skills at three levels, i.e. foundation, intermediate
and advanced [9].
The OpenVM skill set has served as a foundation for constructive alignment in the
design of the set of OpenVM miniMOOCs [15]. The OpenVM skill set includes semantic
definitions of competencies and their cross-references, e.g. “is part of”. The competency
directory provides unique URL access to altogether 33 OpenVM competency defini-
tions [7]. Each miniMOOC addresses one of the eight OpenVM skill areas as learning
outcomes and aligns learning activities and e-assessment to support the achievement
of these learning outcomes. The alignment of learning outcomes, learning activities
and e-assessment has been supported by semantic competency definitions referenced in
the competency directory. Upon successful completion of each miniMOOC at one of
the three levels, a digital micro-credential (Open Badge) can be claimed by the learner
to recognise the specific competency. Digital micro-credentials serve as evidence for
achievement of learning outcomes. The following section describes the current version
of the semantic Competency Directory and its possible uses for Constructive Alignment
in digital learning designs and systems.
Semantic Competency Directory for Constructive Alignment 85
The decentralized semantic web allows to create new competency directories by cross-
referencing the definitions by IRIs. If humans access IRIs, a directory will provide an
HTML website with the competency definition, existing cross-references and a search
interface to navigate the content. If an algorithm fetches IRIs, a directory will answer
with a more suitable format for automatic interpretation like JavaScript Object Notation
Linked Data (JSON-LD) [10]. Alternatively, the answer to a request could always be a
HTML website with embedded semantic data using Rich Structured Data Markup for
Web Documents (RDFa) [11]. Since Open Badge Standard version 1.1 encourages to
use JSON-LD, this format is implemented in the OpenVM project [12]. The vocabulary,
which is important for expressing the semantic meaning of competences and their cross-
references, is the set of noun terms for nodes and relationship terms for edges, which
are used to express the competence definitions and their structure in a graph. To enable
a decentralised network of competence frameworks referencing each other, the ESCO
dataset and vocabulary is used in the OpenVM project. ESCO, the European Skills
Competencies and Occupations Framework, includes vocabulary definitions to express
competences and their relation to each other. ESCO competency directory provides a
web-interface for humans to search and a REST-API that delivers RDF or JSON-LD
data. ESCO enables the expression of semantic relationships within ESCO and to other
competence frameworks [13].
3.2 Requirements
3.3 Implementation
The implementation includes three loosely coupled components, all running in stan-
dalone containers using Docker Runtime. The first component is a Node.js-based back-
end with a REST API (Level 2), which provides the competence entries via unique
IRIs in the JSON-LD format using ESCO vocabulary (Listing 1). It uses the second
component, which is a Neo4j graph database that contains the competency entries as
nodes and the references between them as directed edges. Both, nodes and edges are
tagged with attributes like node-type, reference-type, and available languages. ESCO
was used as vocabulary for node and reference types as far as data allows. The third
component is a web-based user interface created with the React.js framework as a single
page application which fetches information from the REST-API backend (Fig. 1).
Semantic Competency Directory for Constructive Alignment 87
Fig. 1. Screenshot of the frontend for human search with a single-view overlay (entry ID 15)
providing IRIs to be used for deep-linking, e.g. in Open Badge specifications.
The benefits of using the competency directory and adding metadata to the elements of the
learning pathway, e.g. learning outcomes, activities, e-assessment, digital credentials,
can be reached by using the algorithmic ability to recommend suitable resources for
identified skills, adapt learning pathways and define criteria for issuing digital credentials
after passing the e-assessment [8]. Especially in view of the Open Badge Specification
2.0, which allows the use of extended URLs as IRIs, the competency directory may
be used to avoid ambiguities in referencing the same competency in different digital
credentials [15]. The positive effects on better constructive alignment of the learning
design and on improved achievement of the skills by learners need to be evaluated in
user studies in the future. Until end of 2020 the applicability for Open Badge alignment
is planned to be evaluated within the OpenVM project as an expert interview after
integration in the project infrastructure.
88 I. Buchem and J. Konert
Disclaimer: The creation of these resources has been (partially) funded by the ERAS-
MUS+ grant program of the European Union under grant no. 2017-1-DE01-KA203-
003494. Neither the European Commission nor the project’s national funding agency
DAAD are responsible for the content or liable for any losses or damage resulting of the
use of these resources.
References
1. Biggs, J.: Enhancing teaching through constructive alignment. High. Educ. 32(3), 347–364
(1996)
2. Paskevicius, M.: Conceptualizing open educational practices through the lens of constructive
alignment. Open Prax. 9(2), 125–140 (2017). International Council for Open and Distance
Education. https://www.learntechlib.org/p/181424. Accessed 20 June 2019
3. Karunanayaka, S., Naidu, S., Rajendra, J.C.N., Ariadurai, S.A.: Designing continuing profes-
sional development MOOCs to promote the adoption of OER and OEP. Open Prax. 10(2), 179–
190 (2018). International Council for Open and Distance Education https://www.learntechlib.
org/p/183584. Accessed 20 June 2019
4. Boyd, S.: Learning outcomes and opportunities in property education through constructive
alignment. In: Proceedings of the 21st Pacific-Rim Real Estate Society Conference, pp. 1–18
(2015)
5. Remneland Wikhamn, B.: Challenges of adopting constructive alignment in action learning
education. Action Learn. Res. Pract. 14(1), 18–28 (2016)
6. Cross, S., Galley, R.: MOOC Badging and the Learning Arc. OLDS MOOC Blog Post. The
Open University’s repository of research publications and other research outputs (2012).
http://oro.open.ac.uk/42038/7/OLDS%20MOOC.pdf. Accessed 20 June 2019
7. Konert, J., Buchem, I., Lewis, L., Hamilton, G., Riches, T.: Competency alignment of open
badges. In: Stracke, C.M., Shanks, M., Tveiten, O. (eds.) Smart Universities Education’s Digi-
tal Future. Kristiansand, Official Proceedings of the International WLS and LINQ Conference
2017, Logos Verlag Berlin, pp. 30–36 (2017)
8. Buchem, I., et al.: Designing a collaborative learning hub for virtual mobility skills - insights
from the European project Open Virtual Mobility. In: Zaphiris, P., Ioannou, A. (eds.) LCT
2018. LNCS, vol. 10924, pp. 350–375. Springer, Cham (2018). https://doi.org/10.1007/978-
3-319-91743-6_27
9. Firssova, O., Rajagopal, K.: Open VM Competence Framework (2018). https://www.
openvirtualmobility.eu/topics/outputs. Accessed 18 June 2019
10. W3C JSON-LD Working Group: JSON-LD - JSON for Linking Data. https://json-ld.org.
Accessed 19 June 2019
11. The RDF Web Apps Working Group: RDFa. http://rdfa.info. Accessed 20 June 2019
12. IMS Global Learning Consortium: Open Badges v2.0 IMS Final Release. http://www.
imsglobal.org/sites/default/files/Badges/OBv2p0Final/index.html. Accessed 18 June 2019
13. European Commission ESCO handbook. European Skills, Competences, Qualifica-
tions and Occupations (2019). https://ec.europa.eu/esco/portal/document/en/0a89839c-098d-
4e34-846c-54cbd5684d24. Accessed 18 June 2019
14. Konert, J.: O3-A1.2/A1.3 Competency Directory requirements and functional proto-
type (accepted for publication). Open Virtual Mobility Erasmus+ (2019). https://www.
openvirtualmobility.eu/topics/outputs. Accessed 16–18 June 2019
Education Needs in Context of Migration
and Industry 4.0 in Selected EU Countries
1 Introduction
The globalization of migration processes at its best has a positive effect on the economic
development of host countries: it contributes to the growth of gross domestic product,
compensation for a shortage of labor resources, growth and creation of new knowledge,
active development of innovations and others. On the other hand, migration processes
increase the level of competition in local labor markets, reduce the cost of labor and lead
to an increase in unemployment. The study of domestic and foreign experience of studies
devoted to the study of the influence of international labor migration on the development
of labor markets in recipient countries shows that there are many-sided directions for
scientific research. Consistency of regulatory measures is necessary to ensure the safety
and orderliness of migration processes. To reduce or accelerate migration flows, maxi-
mize economic and social preferences while minimizing the costs of migrants and host
communities, you need to know exactly how migration transformations affect economic
policy in the European Union.
Migration has become a key area of the 21st century. National governments are
making great efforts and financial resources to address this problem. The use of ICT in
education is a key role in migration issues. Simultaneously, information and communi-
cation technologies (ICT) have a crucial role in societal progress [1, 2]. The application
of digital technologies in sectors such as governments, nongovernmental organizations
and organic social movements has the potential to improve participation, transparency
and accountability [1, 3].
Several existing studies indicate that other dimensions of education are also helping
to shape return migration. First, studying in a foreign country is a quickly growing
phenomenon around the world (Appave 2010; Boyle et al. 1998; Globerman and Shapiro
2008). The education obtained abroad helps migrants to establish themselves in the
labour market of the host country, but it may be highly valued in the origin country as
well, facilitating return migration. For example, the study by Bijwaard (2010) shows
that most foreign students return to their homeland upon graduating from host country
universities. Second, previous research has established that many immigrants do not find
a job to match their level of education (Hardy 2010), which potentially increases their
willingness to return. Analogous behaviour may be observed when many immigrants
focus on earning the best possible income instead of finding a job that corresponds to
their qualification (Drinkwater et al. 2009; Trevena 2011) [17].
This paper discusses the issue of education migrants and the development of computer
literacy in Czech Republic in context Industry 4.0. Habits of future education and use of
new technology in education are analyzed. Data from Strategic framework for employ-
ment policy by 2030 in the Czech Republic and digital Czech Republic are presented.
Data from Eurostat about migration and education are presented.
The aim of this paper is to analyze the impact of migration processes on the labor
market in the context of Industry 4.0. The impacts of the level of education of individual
migrants from selected countries will be analyzed. Emphasis will be placed on market
requirements within the current fourth industrial revolution and the need for ICT knowl-
edge and skills. The aim of the article is comparison all countries from Visegrad group
and Germany that is the most important country in the export for the Czech economy.
Into deeper evaluation was focused on the main groups of migrants in the Czech Repub-
lic, in Germany connected with Digital Adoption Index and DAI People Sub-index from
countries of their origin.
Literature review encompasses following areas: definitions to key expressions rele-
vant to the scope of the paper like migrants, Industry 4.0, Digital Adoption Index, ICT
and technology enhancing ageing experience.
The article is based on secondary sources. The secondary sources provide informa-
tion about Digital Adoption Index and education, professional literature, information
collected from professional press, web sites, discussions and previous participation at
professional seminars and international conferences related to the chosen subject. It
Education Needs in Context of Migration and Industry 4.0 91
was then necessary to select, classify and update accessible relevant information from
the numerous published materials that provide the basic knowledge about the selected
topic. The DAI is a worldwide index that measures countries’ digital adoption across
three dimensions of the economy: people, government, and business. The index covers
180 countries on a 0–1 scale and emphasizes the “supply-side” of digital adoption to
maximize coverage and simplify theoretical linkages. The overall DAI is the simple
average of three sub-indexes. Each sub-index comprises technologies necessary for the
respective agent to promote development in the digital era: increasing productivity and
accelerating broad-based growth for business, expanding opportunities and improving
welfare for people, and increasing the efficiency and accountability of service delivery
for government [30].
3 Literature Review
International labor migration is an integral part of the global labor market. Being formed
under the influence of the processes of globalization and regionalization, in the condi-
tions of technical and technological development in countries and regions of the world,
modern labor migration is characterized by an unprecedented increasing scale and com-
plexity. With an increase in the population’s spatial mobility, international migration
flows increasingly affect various socio-economic development aspects of host territo-
ries — economic, social, demographic, and other characteristics. On the one hand, the
influx of migrant workers increases the number of labor resources in the territory, and
therefore its economic potential. On the contrary, the competition that arises on the local
labor market, in the absence of large investment projects, stimulates unemployment
growth and reduces wages in industries most dependent on foreign labor. Analysis of
the accumulated domestic and foreign experience in studying the impact of international
labor migration on the development of labor markets in the host territory showed the
breadth and diversity of existing research directions. Given the complex nature of the
labor migration impact on all aspects of territories socio-economic life, there is a need
to develop an integrated economic approach using an index analysis method.
Hedvičáková and Svobodova published papers aimed at labor market and unem-
ployment [4–8], sustainable development [9], Consequences of Industry 4.0 in Business
and Economics [10], indicators focused on well-being, welfare etc. [11] and the educa-
tion of people in the connection with use of advanced technologies [12]. Volejníková’s
published paper, Influence of migration on the labor market in the Czech Republic [13]
and is also focused on unemployment and labor market [14, 15]. Janderová focused
on immigration and asylum policies in the EU [16]. ICT has certainly had a major
impact on the whole process of mobility and migration [19, 20]. In his conceptual study,
Kellerman (2011) concludes that all current mobilities are based on and dependent on
ICT. Whence the emerging research on ICT and it is recognized that ICT have had a
major impact on migration trends by considerably diversifying and increasing migration
opportunities [19, 20]. ICT also directly influence employment; both by creating new
jobs and by acting as a tool that empowers workers to manipulate data and innovate [21,
22]. Telecommunications infrastructure generates employment across a wide range of
sectors and within a variety of professions by aiding in the creation and maintenance of
92 L. Svobodová et al.
3.1 Strategic Framework for Employment Policy by 2030 in the Czech Republic
It appears that the modern labor market requires a more complex approach for some
groups of disadvantaged people, which is not applicable in the current system to the
necessary extent. One of them is further education and retraining. The main problem
with retraining remains, in particular, their ability to respond to labor market demands,
especially in the situations of dynamically changing skill needs. However, employers and
workers are not satisfied with retraining and their process. Therefore, document action
plan of work 4.0 in the Czech Republic was done. Being aware of these new trends,
and the Industry 4.0 Initiative, the Ministry of Labor and Social Affairs in 2016 develop
a research study, Work Initiatives 4.0 (“Work 4.0”). The main effort of the Ministry
of Labor and Social Affairs in this area is to analyze the current situation and future
labor market trends related to the 4th Industrial Revolution and to prepare scenarios
and measures to respond adequately to these changes. Addressing the impact of the 4th
Industrial Revolution, digitization of state administration and the economy on society is a
complex issue and, by its very nature, multiresort with regard to the spheres of social and
economic life that interferes. For this reason, on the basis of Government Resolution No.
629/2018, program Digital Czech Republic was created, which is an umbrella strategic
cross-sectional document. The process of setting up and implementing the program is
coordinated by the Government Council for the Information Society. The Digital Czech
Program consists of three basic pillars: Czech Republic in Digital Europe, Information
Concept of the Czech Republic and The Digital Economy and Society. No less important
role, such as the actual impacts of technological change, will be the demand for some
services and the public sector’s response to it in the future market.
The upcoming topic in the coming decade will be, in particular, to promote the adapt-
ability of workers and their employers to new conditions and to significantly enhance
the role of lifelong learning, including the employment of substantial further education.
Further education will not be influenced only in terms of content or meaning in terms
of acquiring skills and competencies for remaining or retention on the labor market, but
the role of its actors will change as a result of changes in the organization of work.
We can expect to strengthen the role of competences, especially key transferable,
and the emphasis on higher education and skills flexibility and the ability to work with
ICT technologies and related digital literacy [24].
with the advent of the Fourth Industrial Revolution, which is referred to as Industry
4.0 and associated with high use of ICT, labor market conditions vary significantly. It
will also be necessary to educate migrants, when it is again appropriate to use modern
methods of education using ICT. It can be expected that if ICT is used in the home
country and modern technologies are supported, residents will be able to use advanced
technologies more than in countries where they are not used to such a great extent.
For this reason, the 2016 Digital Adoption Index (DAI) was investigated with a second
survey of this indicator. The first was carried out in 2014. Since 2016, newer data have
not been published.
Into comparison were involved firstly all countries from Visegrad group and Ger-
many. Into deeper evaluation was focused on the main groups of migrants in the Czech
Republic, in Germany connected with Digital Adoption Index and DAI People Sub-index
from countries of their origin.
Germany achieved the best results in both Index and Sub-index. The Czech Republic
has achieved similar values with the DAI Business Sub-index as Germany. In the Digital
Adoption Index, the Czech Republic achieved the best result from Visegrad four. The
DAI People’s Sub-Index and DAI Government Sub-Index were similar in the Vise grad
countries but lower than in Germany (Fig. 1).
100%
80%
60%
40%
20%
0%
Czech Republic Germany Hungary Poland Slovak Republic
Fig. 1. Digital adoption index – countries from Visegrad Group and Germany [based on 25]
Based on results of the Digital Adoption Index and DAI People Sub-index from
countries of origin it is necessary to focus on the composition of individual citizenship
in selected states (see Table 1). The Czech Republic and Germany were selected for
comparison. The Czech Republic and Germany are neighboring states [26].
These are economically different countries that differ in size and population too. The
population of the Czech Republic is 10,65 million people and population of Germany is
82,67 million people.
The table above shows that both countries have different migrants with different
nationalities and different percentages. In the Czech Republic, Ukrainians and Slovaks
are the most represented. Until 1993, the Czech Republic was part of Czechoslovakia.
94 L. Svobodová et al.
Table 1. Main country of citizenship and birth of the foreign/ foreign-born population, 1 January
2018 [based on 27]
For this reason, the proportion of Slovaks is so high. In contrast, in Germany, the Turks
are the most represented, followed by Poles, Syrians and Romanians.
This composition of people with different citizenship will have an impact on the
labor market and the ability to learn new ICT skills and knowledge needed for Industry
4.0. Germany responded to the needs of skilled work according to the new immigration
law, skilled labor from abroad with the adequate training and education will face fewer
restrictions when they attempt to get a job in Germany. Any non-EU citizen will now
be permitted to work in Germany if they have the qualified vocational training or degree
course and an employment contract [28].
2018, 2019 and 2020. Scope of the research is an efficient management of migrant inte-
gration requires clear understanding of migrants’ personal and family situation, including
their legal status, origin, cultural background, skills, language skills, medical records,
etc. Once such information is available to public authorities, it can improve societal
outcomes to the benefit of both host countries and migrants.
Expected Impact are new or enhanced ICT solutions and tools will facilitate the
efforts of public administrations at EU, national and local levels to manage the integration
of migrants. They will allow for developing and deploying the necessary processes and
services in the view of the efficient identification and inclusion of migrants. They will also
facilitate communication with migrants and their access to services such as community
language teaching, education, training, employment, welfare and healthcare systems
within the host communities [30]. The need for continuous training and retraining will
become increasingly important in the years to come. The more national governments are
able to use modern ICT in this area thus gain greater competitiveness and will reduce
the costs associated with rising levels of unemployment.
Interesting is the experience with teaching migrants and inclusive education from
Sweden. This relates to special education, as attainment is central to the definition of the
need for special educational support in Sweden. Similar to international patterns, pupils
with lower socio-economic background and with migrant-backgrounds are overrepre-
sented among SEN-pupils in Sweden (Giota and Lundborg 2007; Berhanu 2008; Dyson
and Berhanu 2012; Dyson and Berhanu 2012; Cook and Kiru 2018; Cook and Kiru
2018), groups that exercise choice to a lower degree (Bunar 2010; Daun 2003). Pupils
from different socio-economic backgrounds are thus likely to experience different choice
situations and to experience different outcomes from their choices (Vamstad 2014) [31].
Docquier et al. (2012) investigate the relationship between remittances and migrants’
education both theoretically and empirically, using original bilateral remittance data. At
a theoretical level they lay out a model of remittances interacting migrants’ human capital
with two dimensions of immigration policy: restrictiveness, and selectivity. The model
predicts that the relationship between remittances and migrants’ education is ambiguous
and depends on the immigration policy conducted at destination. The effect of education
is more likely to be positive when the immigration policy is more restrictive and less
skill-selective [18].
where they are not used to such a great extent. It will also be possible to better educate
them with modern education methods (e.g. blended learning or eLearning, etc.) For this
reason, Digital Adoption Index and DAI People Sub-index from countries of origin were
used in the article after comparing digital progress.
Highly skilled workers are key drivers in the contemporary knowledge-based econ-
omy with destination countries making increasing efforts to attract immigrants from this
group, while emigration countries are equally attempting to encourage them to move
back home (Beine et al. 2001; De Haas 2010; Jakoby 2011; Stark et al. 1997; Thaut
2009). Perhaps the most easily accessible variable, describing “skills”, is education.
Previous research on the relationship between the level of education and return migra-
tion has presented mixed evidence. Based on Swedish data, Nekby (2006) found that
returning emigrants have higher levels of education compared to those who stay, i.e.
the initial “brain drain” could become a “brain gain” for the source country. Jensen
and Pedersen (2007) obtained a similar result for all immigrants leaving Denmark, but
their findings were less straightforward by source country groups. In contrast, Dustmann
(1996; 2003) found that there was a negative effect of years of schooling on the intention
of immigrants living in Germany to return to their home countries. These ambiguous
results call for a more comprehensive treatment of education together with an analysis
on the association between skills and return migration behavior [17].
In further research we will deal with the economic impact of migration on the Czech
economy. Work 4.0 was developed for the ongoing Industry 4.0 initiative, which specifies
the market needs for Industry 4.0. However, when examining the effects of migration
on the economy, two key aspects of the component need to be taken into account: the
Czech Republic has the lowest unemployment rate across the EU (2% in June 2019) and
an aging population. On the Czech labor market there is an excess of demand and supply
of labor and companies are not able to execute all the orders they have, which results
in a slower GDP growth (2.7% in the second quarter of 2019). The second area to be
examined will be the calculation of the number of migrants needed annually to ensure
the stability of the social and pension system due to the aging of the Czech population.
Acknowledgement. The paper is supported by the project SPEV 2019 at the Faculty of Infor-
matics and Management of the University of Hradec Kralove, Czech Republic. In addition, the
authors thank Anna Borkovcova for her help with the project.
References
1. Altenburg, F., Faustmann, A., Pfeffer, T., Skrivanek, I.: Migration und Globalisierung in Zeiten
des Umbruchs. Festschrift für Gudrun Biffl. Krems (Edition Donau-Universität Krems),
pp. 417–434 (2017)
2. Majchrzak, A., Markus, M.L., Warcham, J.: Designing for digital transformation: lessons for
information systems research from the study of ICT and societal challenges. MIS Q. 40(2),
267–277 (2016)
3. Majchrzak, A., Markus, M.L., Warcham, J.: Call for papers. MISQ special issue on ICT and
Societal Challenges (2012)
Education Needs in Context of Migration and Industry 4.0 97
4. Hedvicakova, M., Svobodova, L.: Impact of low unemployment on the labour market in the
context of the Industry 4.0. In: Vision 2020: Sustainable Economic Development, Innovation
Management, and Global Growth, vol. I–IX, pp. 5450–5455. Madrid (2017) (2017)
5. Hedvicakova, M., Svobodova, L.: The labour market of the Czech Republic in the context
Industry 4.0. In: 20th International Colloquium on Regional Sciences Location. Kurdejov,
pp. 303–310 (2017)
6. Hedvicakova, M., Svobodova, L.: Unemployment in the European Union with the empha-
sis on the Visegrad Four. In: Vision 2020: Sustainable Economic Development, Innovation
Management, and Global Growth, vol. I–VII. Seville, pp. 4217–4221 (2016)
7. Hedvicakova, M., Sokolova, M., Mohelska, H.: The impact of economic growth on wages
and the supply of employee benefits in the Czech Republic. Transform. Bus. Econ. 17(3),
140–154 (2018)
8. Hedvicakova, M.: Unemployment and effects of the first work experience of university
graduates on their idea of a job. Appl. Econ. 50(31), 3357–3363 (2018)
9. Bednarska-Olejniczak, D., Olejniczak, J., Svobodova, L.: Towards a smart and sustainable
city with the involvement of public participation-the case of Wroclaw. Sustainability 11(2),
332 (2019)
10. Maresova, P., et al.: Consequences of Industry 4.0 in business and economics. Economies
6(3), 46 (2018)
11. Svobodová, L., Černá, M., Hruša, P.: Selected simple indicators in the field of advanced
technologies as a support of SMART cities and their impact on tourism. In: Kar, A., et al.
(eds.) Digital Nations – Smart Cities, Innovation, and Sustainability. I3E 2017. Lecture Notes
in Computer Science, vol. 10595, pp. 172–182. Springer, Cham https://doi.org/10.1007/978-
3-319-68557-1_16
12. Svobodova, L., Hedvicakova, M.: Use of smart technologies for hybrid learning as a way to
educate people became full smart cities residents. In: Nguyen, N., Pimenidis, E., Khan, Z.,
Trawiński, B. (eds.) Computational Collective Intelligence. ICCCI 2018. LNCS, vol. 11056,
pp. 419–428. Springer, Cham (2018). https://doi.org/10.1007/978-3-319-98446-9_39
13. Volejnikova, J., Knezackova, R.: Influence of migration on the labour market in the Czech
Republic. In: 12th International Scientific Conference on Economic Policy in the European
Union Member Countries, Ostravice, pp. 892–900 (2015)
14. Volejníková, J.: Structural unemployment in the Czech Republic. In: 13th International
Scientific Conference on Hradec Economic Days, Hradec Králové, pp. 343–349 (2015)
15. Knezackova, R., Volejnikova, J.: The labour market in the Czech Republic after accession to
the European Union. In: 8th International Days of Statistics and Economics, Praha, pp. 687–
697 (2014)
16. Janderova, J., Gyamfi, S.: Common European asylum system evolution and its perspectives.
In: 5th International Multidisciplinary Scientific Conference on Social Sciences and Arts
SGEM 2018, Vídeň, pp. 235–244 (2018)
17. Pungas, E., Toomet, O., Tammaru, T., Anniste, K.: Are better educated migrants returning?
Evidence from multi-dimensional education data, Norface migration discussion paper no.
2012–18 (2019). http://ww.w.norface-migration.org/publ_uploads/ndp_18_12.pdf
18. Docquier, F., Rappoport, H., Salomone, S.: Remittances, migrants’ education and immigration
policy: theory and evidence from bilateral data. Reg. Sci. Urban Econ. 42(5), 817–828 (2012)
19. Codagnone, C., Kluzer, S.: ICT for the Social and Economic Integration of Migrants into
Europe. Luxembourg: Office for Official Publications of the European Communities (2011)
20. Collin, S., Karsenti, T.: ICT and migration: a conceptual framework of ICT use by
migrants. In: Conference: World Conference on Educational Multimedia, Hypermedia and
Telecommunications At: Denver, Colorado, USA, vol. 2012, pp. 1492–1497 (2012)
98 L. Svobodová et al.
21. Raja, S., Imaizumi, S., Kelly, T., Marimatsu, J., Paradi-Guilford, C.: Connecting to Work: How
information and communication technologies could help expand employment opportunities.
(Report – ICT Sector Unit, The World Bank) (2013)
22. Datta, A., Agarwal, S.S.: Telecommunications and economic growth: a panel date approach.
Appl. Econ. 36(15), 1649–1654 (2004)
23. Andrianaivo, M., Kpodar, K.R.: ICT Financial Inclusion and Growth: Evidence from African
Coutries (IMF Working Paper WP/11/73, International Monetary Fund) (2011)
24. Strategic framework for employment policy by 2030 in the Czech Republic
25. World Bank, Digital Adoption Index. http://www.worldbank.org/en/publication/wdr2016/
Digital-Adoption-Index
26. Svobodova, L., Hedvicakova, M.: Technological readiness of the Czech Republic and the
use of technology. In: Themistocleous, M., Morabito, V. (eds.) Information Systems. EMCIS
2017. Lecture Notes in Business Information Processing, vol. 299 (2017). Springer, Cham.
https://doi.org/10.1007/978-3-319-65930-5_53
27. Eurostat, Main countries of citizenship and birth of the foreign foreign born. https://ec.
europa.eu/eurostat/statistics-ex-plained/images/f/f6/Main_countries_of_citizenship_and_
birth_of_the_foreign_foreign-born_population%2C_1_January_2018_%28in_absolute_
numbers_and_as_a_percentage_of_the_total_foreign_foreign-born_population%29.png
28. Germany Visa. Germany’s new immigration laws open door for skilled labor. www.germany-
visa.org/germanys-new-immigration-laws-open-door-for-skilled-labor/
29. Bertoli, S. Brücker, H.: Selective immigration policies, migrants’ education and welfare at
origin. Econ. Lett. 113(1), 19–22 (2011)
30. European Commission, Horizon (2020). https://ec.europa.eu/info/funding-tenders/
opportunities/portal/screen/opportunities/topic-details/dt-migration-06-2018-2019
31. Magnússon, G.: Inclusive education and school choice lessons from Sweden. Eur. J. Spec.
Needs Educ. 35(1), 25–39 (2019). https://doi.org/10.1080/08856257.2019.1603601
Students’ Language Needs Analysis
as a Motivation Mover - Czech and Taiwanese
Case Study
1 Introduction
The paper discusses a proved pedagogical scenario. Students’ needs analysis is com-
monly used as a starting point for the teaching/learning process especially in the uni-
versity environment [1]. Affective learning dominates in this scenario with its affective
objectives like attitudes, feelings and motivation [2]. Currently technology enhanced
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 99–108, 2020.
https://doi.org/10.1007/978-3-030-38778-5_13
100 M. Cerna and C.-J. Lin
environment is characterized by involvement of social media and other Web 2.0 tools
into the process of education, into creation, adaptation and sharing of study content mate-
rial together with evaluation of students’ performance within blended learning concept
[3–6].
The authors approached the case study from students’ perspective; gained findings
belong to the category of pedagogical experience with e-learning. The paper contributes
to blended instruction of online and traditional approaches based on students’ experience
and needs.
This study investigates students’ language learning behavior, their approach and
motivation in Czech and Taiwanese university environment. The paper focuses on stu-
dents’ experience in utilization of online study material and correlation between their
needs in practicing individual language competences and preferable form of study mate-
rial and running of the teaching/learning process in mastering the English language as a
second language.
A mixed qualitative-quantitative research was conducted. Data collection consisted
of the survey based on the key research tool, which was a questionnaire on Students’
language needs and a follow up-discussion on students’ experience, expectations and
perceived potential benefits of on-line sources as well as traditional approaches.
The output of the study brought the following conclusions. The case study has shown
that students’ learning background, experience and behavior was different in several ana-
lyzed areas in two compared environments but highly inspiring and leading to the fruitful
discussion enlarged with an intercultural dimension. There were statistically significant
differences between not only the two samples but between male and female students, as
well. The tutors’ interest in students’ needs proved to be a motivating factor enabling
students to think of their language needs, formulate them and discuss them so that
they could adapt and widen their language learning habits. The awareness of students’
needs and students’ behavior in utilizing virtual space, including social applications,
websites, films, videos or podcasts has a great potential to provide fruitful learning
environment [5].
2 Methodological Frame
This chapter provides readers with literature review, state of art, goals and individual
stages of the research, research tool and sample.
Web 2.0 and incorporation of its tools into the process of education at the Faculty
of Informatics and Management, University of Hradec Kralove has been part of its
policy since its establishment 25 years ago. Utilization of the Internet and eLearning
experience on the wide scale of all students entering the Faculty of Informatics and
Management has been systematically monitored for more than two decades, see more
in the longitudinal study [14]. The academics approach the research on the Internet
use from various perspectives. As for language teaching/learning, Kostolanyova and
Simonova explored development of language competences and designed an adaptive
model of e-Learning where learner’s sensory characteristics were reflected [15]. Cerna
focused on the power of motivation, she applied in her research on the development of
language competences a psychodidactic approach, and designed a modified expectancy
model [1].
In this study, we focus on students’ learning behavior, their approach to learning
English language as a second language, their experience with utilization of virtual space
as well as traditional ways of learning. The survey itself has a decade long history; it has
been repeatedly run at the beginning of academic year with both full-time and part-time
students of Information Management, Financial Management and Applied Informatics
within the frame of bachelor study programmes at the University of Hradec Kralove
(UHK) [5]. In May 2019 the survey was newly conducted during Erasmus+Mobility
Programme at the National Taiwan University of Science and Technology in Taipei,
Graduate Institute of Digital Learning and Education with students attending classes of
Language learning with digital technology and classes of Motivation and Learning.
102 M. Cerna and C.-J. Lin
The goal of the paper is to present findings relating to current language needs of stu-
dents, their learning behavior and their experience in studying languages with focus on
utilization of Web 2.0 tools in two university settings.
• The first sub-goal is to compare selected findings and highlight similarities and
differences between students from different language and cultural backgrounds.
• The other sub-goal refers to students’ language competences measured according to
CEFR scale. The sub-goal is to put into context reached levels of students’ language
competences with students’ requirements on desirable improvement of their language
skills.
The analyzed questions were following: Q1 – Do you study ‘on the Internet’? Q2 –
What language educational websites have you used for practicing languages? Q3 – Which
of these websites would you recommend and why? Q4 – Which functionalities should
the language educational web-site have to fit your requirements? Q5 – What would you
like to practice most during ‘our’ language classes? Q6 – What is the biggest trouble:
Listening, reading, writing, grammar or just speaking? Q7 – What level of English are
you at A1 – C2? Q8 – What fits you when you are studying languages? – Internet website,
language e-course, F2F classes, textbook, films, taking notes or keeping a log.
There were 45 students in each group in UHK and NTUST sample of students, who
submitted the questionnaire. All participants studied English language as their second
language. Their language competence was comparable as can be seen in the subchapter
Findings 3.4. The UHK sample consisted of Czech students attending bachelor study
programme at the Faculty of Informatics and Management, their age span was 20 to 24.
The NTUST sample consisted of students from Taiwan, Malaysia, Vietnam, Thailand,
two students were from Kenya and two students were from Ethiopia. Students attended
Graduate Institute of Digital Learning and Education. Their age span was also 20 to 24,
four students were over 30, those were students from Ph.D. programme.
3 Findings
Only 7% of students from the NTUST sample don’t use the Internet for language study
purposes. In UHK sample there are astonishing 25% of students who claim that they do
not use the Internet for studying English language. There are big differences between
men and women in the Internet use in the discussed issue: 91% of men to 76% of women
at NTUST, 59% of men to 75% of women at UHK. From the follow up discussion we
learned that women in the UHK sample prepare more regularly for classes, e.g., they do
the tasks in Blackboard and use other on-line sources like BBC learning English. High
percentage of male students start using on-line sources at the end of semester when they
have to sit for listening and reading tests which are based on sources from language
portals.
3.2 Q2 - What Language Educational Websites Have You Used for Practising
Languages?
The second question was focused on students’ experience with language websites and
applications. Findings are illustrated in the Fig. 1.
Website ‘Help for English’ dominates in the UHK student sample; nearly half of
students knows and uses this language educational portal. Popularity of the language
portal is especially based on the fact that it is a Czech portal, it means students are given
explanations in Czech, more over students know it from secondary school.
As for the sample from Taiwan-tech, there was no similar portal mentioned, no one
used any language educational portal at the secondary school.
104 M. Cerna and C.-J. Lin
The most often mentioned sites in Taiwan-tech student sample is YouTube reaching
30%. ‘YouTube’ used to be the most favorite social application in former surveys [5].
But in the latest survey it was mentioned only 4 times out of 45 UHK students.
BBC is a highly represented web in both samples. BBC Learning English is globally
well-established language web portal with abundant functionalities and links to main
social nets reflecting IT and social development, changes and demands. But the ratio of
current UHK users is getting lower. BBC learning English websites are used by a quarter
of Taiwan-tech.
Wide popularity still holds the ‘Duolingo’ application in UHK student sample, in
previous years those were mostly part-time students who used this application. Only one
student from Taiwan-tech University uses this application. During a follow-up discus-
sion, this application had success; students may enrich their used language sources with
this app.
A new player entered the scene of utilized educational language web sites by UHK
student sample, which is a portal Perfect English Grammar. Taiwan-tech students noted
down 3 times ‘Grammarly.com’ sites 3 times.
20% of Taiwan-tech students use VoiceTube sites, which are unknown to UHK
students. Another unknown site which is popular in Taiwan-tech students is ‘Ted Talks’.
3.3 Q5 and Q6 - What Would You like to Practice Most During Language
Classes? What Is the Biggest Trouble: Listening, Reading, Writing,
Grammar or just Speaking?
Due to the place limitation findings relating required functionalities of websites are
not presented in this paper. Results from Q5 and Q6 are visualized in one graph and
compared as illustrated in Fig. 2. The blue bar represents: what students would like to
practice most and the orange one what causes them the biggest troubles.
In both samples practising of language speaking skill dominates. Comparable results
refer to practising speaking.
UHK sample shows slightly more balanced results in practising skills and individual
segments of language.
Students’ Language Needs Analysis 105
Fig. 2. What do the students want and need to practice most? (Color figure online)
More than 70% NTUST students want to practice speaking during classes and over
a half of them realizes or feels that they have problem with this skill.
Nearly half of the UHK respondents feels that they have troubles with grammar and
would like to learn it during classes as well as in the e-course. There can be seen big
difference between two samples in their approach to grammar. In the NTUST sample
only every fifth student realizes that grammar might be a problem but they do not require
grammar issue to be incorporated into language classes.
Grammar and Writing skill represent key differences between two samples and shows
one of few differences caused by students’ background.
Next findings come from the follow-up discussion with students. Czech students
insist more on particulars, they focus on prepositions, articles and somehow it makes
a problem for many of them to grasp the whole unit – e.g. write an essay or take the
minutes. Moreover, they believe that ‘Mr. Google’ will do their job in sense of writing
and checking their written tasks. The NTUST students realize the importance of good
level of writing skills. At NTUST University, they have to submit a great deal of written
assignments. One third of them realizes that they have problem with this skill but only
four of them would like to incorporate development of this skill into the classes.
writing skills in NTSUT students, especially of those with the higher and highest level of
competences feel writing skill as a problem. On the other side in UHK students grammar
and speaking dominate. Students of lower language competence focus on speaking, lis-
tening so that they could be able to understand and start communicate. Students reaching
higher levels of language competence include grammar category more frequently to be
practiced. They do not have problems with understanding, they can focus on mastering
the language, where use of proper grammar is desired.
where they believe ‘everything’ can be explained without their effort. This conclusion is
based on findings from both questionnaire as well as the follow-up discussion. None from
Taiwan-Tech respondents mentioned games or songs in classes during the discussion.
Rather surprising fact is that UHK participants favored language websites so little. In
spite of the fact that they gained some experience with websites from previous years but
they prefer more entertaining way of language study. As for language websites, students
from Kenya made an interesting note. They said: “we don’t need to speak perfect and
study systematically from websites, we are from multilingual environment. We need to
speak, we need to communicate, we don’t need to be perfect.” UHK sample is strongly
influenced by Czech environment where shyness to speak plays an important role. Fear
of making mistakes greatly limits communication.
4 Conclusions
The goal and sub-goals of the paper were reached. The comparative study brought an
updated view of the current use of sources supporting practicing language in Czech and
Taiwanese university settings. The study shows difference in priorities in utilization of
study materials which reflects different language approaches stemming from different
cultural background of the compared groups.
As a conclusion, it can be stated that conducting language needs analysis with a
follow up discussion and consequently with worked out electronic presentation of find-
ings presented and shared on the virtual learning platform is a proven pedagogical
scenario applicable in various setting; bringing updated, beneficial and inspiring ideas
to all ‘stakeholders’. In this scenario both teachers and students profiteer: teachers can
adapt study materials to current situation based on formulated requirements, experience
and expectations and students get motivated as they can see that their needs are dis-
cussed, taken into consideration in further planning, respected and shared. Intercultural
dimension represents precious added value.
Acknowledgements. The paper is supported by the project SPEV 2019 at the Faculty of Infor-
matics and Management of the University of Hradec Kralove, Czech Republic. In addition, the
authors thank Anna Borkovcova for her help with the project.
References
1. Cerna, M.: Psychodidactic approach in the development of language competences in univer-
sity students within blended learning. Open Learn. J. Open Distance e-Learn. 33(2), 142–154
(2018). https://doi.org/10.1080/02680513.2018.1454834
2. Mohanty, A.: Cognition, Affection and Conation: Implications for Pedagogical Issues in
Higher Education (2010). pedagogy.iitkgp.ernet.in/downloads/Cognition_Pedagogy.doc
3. Greener, S.: How are web 2.0 technologies affecting the academic roles in higher educa-
tion? A view from the literature. In: 11th European Conference on e-Learning, ECEL 2012,
Groningen, pp. 124–132 (2012)
4. Klimova, B., Poulova, P.: Pedagogical principles of the implementation of social networks at
schools. In: Gong, Z., Chiu, D.K.W., Zou, D. (eds.) ICWL 2015. LNCS, vol. 9584, pp. 23–30.
Springer, Cham (2016). https://doi.org/10.1007/978-3-319-32865-2_3
108 M. Cerna and C.-J. Lin
5. Cerna, M., Borkovcova, A.: Enhancing students’ involvement in the process of education
through social applications. In: Uskov, V.L., Howlett, R.J., Jain, L.C., Vlacic, L. (eds.) KES
SEEL-18 2018. SIST, vol. 99, pp. 185–193. Springer, Cham (2019). https://doi.org/10.1007/
978-3-319-92363-5_17
6. Popescu, E., Kubincová, Z., Homola, M.: Blogging activities in higher education: comparing
learning scenarios in multiple course experiences. In: Li, F.W.B., Klamma, R., Laanpere, M.,
Zhang, J., Manjón, B.F., Lau, R.W.H. (eds.) ICWL 2015. LNCS, vol. 9412, pp. 197–207.
Springer, Cham (2015). https://doi.org/10.1007/978-3-319-25515-6_18
7. Valtonen, T., et al.: Net generation at social software: challenging assumptions, clarifying
relationships and raising implications for learning. Int. J. Educ. Res. 49, 210–219 (2010)
8. Cerna, M., Poulova, P.: Social software applications and their role in the process of education
from the perspective of university students. In: Proceedings of the 11th European Conference
on e-Learning, pp. 87–96. Academic Publishing, Groningen, Reading (2012)
9. Lin, Ch. J.: An online peer assessment approach to supporting mind-mapping flipped learning
activities for college English writing courses. J. Comput. Educ. Springer, Heidelberg (2019).
https://doi.org/10.1007/s40692-019-00144-6
10. Smart, K.L., Cappel, J.J.: Students’ perceptions of online learning: a comparative study. J.
Inf. Technol. Educ. 5, 201–219 (2006)
11. Tess, P.A.: The role of social media in higher education classes (real and virtual) – a literature
review. Comput. Hum. Behav. 29, A60–A68 (2013)
12. Schroeder, A., Minocha, S., Schneider, C.: The strengths, weaknesses, opportunities and
threats of using social software in higher and further education teaching and learning. J.
Comput. Assist. Learn. 26, 159–174 (2010)
13. Cheung, S.K.S.: Distance-learning students’ perception on the usefulness of open educational
resources. In: Cheung, S.K.S., Kwok, L., Ma, W.W.K., Lee, L.-K., Yang, H. (eds.) ICBL 2017.
LNCS, vol. 10309, pp. 389–399. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-
59360-9_34
14. Poulova, P., Cerna, M.: Utilization of the internet and eLearning experience in students enter-
ing university-a longitudinal study. Adv. Sci. Lett. 24(4), 2573–2577 (2018). https://doi.org/
10.1166/asl.2018.11008
15. Kostolanyova, K., Simonova, I.: Learning English through the adaptive model of e-learning
reflecting learner’s sensory characteristics. In: Cheung, S.K.S., Kwok, L., Kubota, K., Lee,
L.-K., Tokito, J. (eds.) ICBL 2018. LNCS, vol. 10949, pp. 57–68. Springer, Cham (2018).
https://doi.org/10.1007/978-3-319-94505-7_4
16. Brannen, J.: Mixing Methods: Qualitative and Quantitative Research. Routledge, London
(1992). https://doi.org/10.4324/9781315248813
Uncovering the Potential of the Google Search
Engine for L2 Learning and L2 Translator
Training
David Mraček(B)
University of Hradec Králové, Rokitanského 62/26, 500 03 Hradec Králové, Czech Republic
david.mracek@uhk.cz
Abstract. The paper presents a study investigating the use of the Google Search
Engine in L2 translation and L2 learning. A group of undergraduate students in
Tourism Management was introduced to three tips for effective Google search as
part of their optional course in the basics of translation. A week later, half of the
students attended a session devoted to further practice of the tips. Later, both groups
were asked to translate a short text from Czech, their native language, into English
with the assistance of the Google Search Engine. Drawing on the participants’
translations, written protocols and search histories, the study analyses their online
behavior during L2 text production, providing examples of successful applications
of search strategies and explaining cases where online search techniques failed to
connect effectively with the user’s language skills and reflection.
1 Introduction
The Google Search Engine (GSE) has become a permanent presence in people’s daily
lives and the default option that people refer to if they have a query about a certain topic,
evidenced by the fact that it processes an average of around 40,000 search queries per
second [1]. However, its benefits for language learning and translation are much less
known and researched.
The present paper explores several of many ways in which the GSE can be exploited
by users of a foreign language (here referred to as L2), be it learners, translators, or in fact
any text producers. It presents the methodology and results of a study which was a follow-
up on an earlier investigation into the opportunities that the GSE offers to L2 translation
teaching [2]. And while the previous study suggested that the tool can, in general, improve
the quality of written language and enrich the learning experience, some questions were
left unanswered and several limitations were identified in the methodology. The aim of
the present study was, therefore, to place more emphasis on the search process itself in
an attempt to shed more light on the possibilities and limitations of the GSE in respect
to its use during L2 text production.
2 Methodology
As the present study is intended as a contribution to the author’s long-term research into
the possibilities of the GSE in L2 learning and translator training, its methodology is
inspired by his previous investigation [2]. The aim is to explore how the tool can assist
the language user in producing better L2 material, specifically more acceptable grammar
and lexicon. The focus is on the following three basic search strategies (the examples
are taken from [2]):
(1) enclosing the query (a string of words) in quotes makes the GSE return results with
those exact words in that exact order; this operator is a sine qua non of effective
Google search and will be referred to as the quotes operator throughout the present
text; e.g. phrases “According to a recent study” or “it will benefit the elderly” all
yield a high number of hits;
(2) doing a wildcard search by using an asterisk (*) in the middle of a string of words
(which is itself enclosed in quotes), with the asterisk acting as a placeholder for one
or more words; the mechanism will be referred to as the asterisk operator; e.g. the
query “countries have * measures to” produces results including the verbs taken,
adopted and introduced;
(3) restricting the search (of a phrase enclosed in quotes) to a specific domain by using
the site operator; e.g. the phrase “suffer unwanted side effects” can be restricted
to US institutions of higher education by adding site:edu or to British websites by
typing site:uk into the search box after the query enclosed in quotes (with one space
between them).
At the same time, three of the limitations identified in the previous study were
taken into consideration in the methodology design. Firstly, as the previous experiment
included only one in-class demonstration of what appears to be a largely new set of
skills, the question emerged whether exposing the students to a further demonstration
and practice session might help them better understand the possibilities offered to them
by the GSE. Secondly, the time period between the in-class demonstration and the
submission of the home assignment was reduced substantially compared to the previous
experiment to ensure that the know-how demonstrated to the students is not forgotten
by the time they can utilize it in the home assignment. Thirdly, to cast more light on
the actual role of the GSE in the process of L2 production, the data collection has been
made more nuanced so that the GSE can be isolated from other tools and strategies used
by the user or, alternatively, its interaction with these other tools and strategies can be
better explained. Therefore, a combination of data collection methods more conducive
to a detailed qualitative analysis was selected for the present study; similarly, the small
number of participants in the present study enabled more thorough investigation.
The study involved four students enrolled in their second year of a bachelor’s degree
program in Tourism Management at the University of Hradec Králové, Czech Republic,
with their English language skills between levels B2 and C1 of the Common European
Framework of Reference for Languages. The optional course in translation, of which
the experiment was part, aims at developing, inter alia, students’ L2 writing competence
and research skills, making the GSE an ideal educational solution.
Uncovering the Potential of the Google Search Engine 111
The experiment consisted of two stages, described in detail in the subsections below:
(1) two in-class Google Search Engine demonstrations, (2) home assignment comprising
the translation of a short text and the completion of a verbal protocol.
One of the translation seminars in April 2019 was devoted to introducing the students to
three of the numerous tips for effective use of the GSE, namely (1) the quotes operator,
(2) the asterisk operator, and (3) the site operator. The present author believes these three
tips to be well applicable to translation and foreign-language learning.
During the demonstration, the teacher (i.e. the present author) explained the three
tips and conducted a number of GSE searches, using his own examples, on the classroom
computer, with everything projected onto a screen at the front of the room. To promote
active learning, the students were encouraged to follow the teacher step by step using
their own laptops, which they had been asked to bring.
As the previous study [2] has shown, these tips are rather unknown even among
students in their early twenties, who are otherwise known to be conversant with the
latest technology and accustomed to using the GSE for private and educational purposes
on a regular basis. Therefore, a second demonstration was held to help the students
better internalize the tips that had been explained to them. As a preparatory exercise, the
students were asked to translate a short text from Czech into English at home and bring
their translations to the class the following week. The text, an extract of 75 words from a
newspaper article on the overuse of medication by elderly Czechs, was of similar length
and difficulty and contained a similar range of linguistic issues to the one the students
were to translate in the next stage of the experiment (see Sect. 2.2).
The second demonstration, which was only attended by two students, revolved
around the translation assigned for home practice. The teacher and the students dis-
cussed the text phrase by phrase, sentence by sentence, commenting on the three GSE
search tips – and various combinations thereof – and how they could assist the user in
producing a good English translation text. Collectively, various search scenarios were
tested, using the classroom projector and the students’ own laptops. Although the pri-
mary, narrow focus was on the correct use of articles, prepositions and collocations, the
more general objective was to enable the students to better understand the complexity of
the translation process as well as the possibilities and limitations of the GSE in L2 text
production; most importantly, as Mraček [2] says, students need to learn to accept and
appreciate that “items that can be entered into the search box go well beyond isolated
words, and can include entire clauses or even sentences.”
as best as they could; they were told that their translations would be used in a study on
the GSE but would not affect their overall course assessment.
The source text was a short extract (with some minor editing) from a news item
on the rising use of prescription drugs among the elderly published in a Czech daily.
Although the text is composed of several complex sentences, it is easy to understand
for non-professionals and contains general vocabulary, being an efficient instrument for
testing students’ grasp of grammar rules (e.g. word order, articles, countability of nouns,
verb tenses) and use of English in general (collocations, prepositions).
The following is the text in Czech:
The number of people in the United States who take a multivitamin pill every
day has gone up by ten percent over the last twenty years, and the US is now the
world leader in the consumption of diet supplements. Doctors’ experience suggests
that the Czechs are going down the same path. The problem, scientists say, is that
people believe they cannot do them any harm. But if too many of these supplements
accumulate in the body they can be fatal. The experts all agree on one thing: if
diet supplements were only available on prescription, it would reduce the risk of
people overusing them.
3 Data Analysis
The following section contains the analysis of the data obtained through the three meth-
ods described above. Each subsection deals with one participant, describing his/her GSE
behavior and how it relates to the quality of the translation. Each participant’s translation
is followed by a quantitative summary of the search process (stating the total number of
queries, proportion of searches involving one of the three operators) and a discussion of
selected searches, with special focus on the length and structure of the query and how
effective it was in assisting the user to reach an acceptable solution; examples in the
translations are numbered using superscript. The intention is not to conduct a thorough
assessment of the translations in terms of language quality or translation procedures
used. Rather, the objective of the analysis is to attempt at identifying the profile of
each participant with regard to his/her behavior on the GSE and how it may be affected
by whether he/she took part in two GSE sessions (participants 1 and 2) or only one
(participants 3 and 4).
The one search which contained no operator (the united states is or are, Ex. 1) helped
the translator access an article on whether the name of the country should be treated as
singular or plural, ultimately leading her to the correct conclusion.
A total of eight commentaries were made in the verbal protocol and all reflect good
search strategies which allow the translator to produce acceptable English. Besides Ex.
1, examples of other effective searches include the following:
Ex. 2: “people in the United States” site:edu vs. “people in United States” site:edu
Ex. 3: “people who swallow” site:uk vs. “people which swallow” site:uk
Ex. 4: “grown by 10%” site:uk to check the preposition which normally follows the
verb (moreover, a noun phrase was included in what is a laudable attempt to verify an
entire part of a sentence)
Ex. 5: “in the last twenty years” site:uk vs. “in last twenty years” site:uk. When looking
at the number of hits returned for either version, the translator correctly concluded that
the former (with the definite article) is to be preferred.
As the examples above show, the translator was not afraid to make multi-word
queries involving entire parts of sentences (complex adverbials etc.), frequently using
the combination of quotes and the site operator to ensure the excerpts returned came from
British and US sites only (increasing native-speaker authenticity), with queries mostly
aimed at comparing two versions offered to the user by her intuition (nouns with/without
an article, two alternative verb forms etc.).
However, except for Ex. 6 (where an asterisk was used as a placeholder for the prepo-
sition “experts agree * one thing”), the searches were confined to the first two sentences.
It can be argued that if the translator had applied the above strategies throughout the
entire translation process, she could have avoided a number of infelicities found in the
rest of the text. For instance, the phrase doctor’s experience show (Ex. 7) contains two
mistakes (doctors’ and experience shows are correct); a more critical approach to the
translator’s own knowledge of grammar (which she did manifest when verifying such
basic issues as people who/which, see Ex. 3) could have led her to review this particular
formulation. Similarly, spelling or typographic errors such as those in Examples 8, 9,
10, 11 can be avoided, with the GSE drawing the user’s attention to them; punctuation
rules (Ex. 12: however is normally followed by a comma when placed initially) can be
inferred by the user from the excerpts returned by the GSE providing the word is part of
a longer query (e.g. “however, the number” site:edu).
In sum, participant No. 1, who attended both GSE demonstrations, manifested a
solid grasp of the know-how presented to her, which enabled her to be in better control
of grammar issues such as articles, pronouns, prepositions and verb forms. She failed,
however, to maintain the same level of rigor till the end of the exercise, which may have
been due to lack of time or other factors unknown to the author.
worldwide in3 using food supplements. Doctors’ experience show, that even Czech people
follow the same way. According to scientists the problem is that many people think food
supplements can’t cause any damages. But if there’s too much5 food supplements in
the body, they can cause even death. Professionals agree on one thing: if the food
supplements were on prescription, the risk of their overuse would decrease.
The translator did a total of 83 searches, 82 of which using quotes (the one search
where no quotes were used was immediately followed by a repeated query with quotes
as the translator realized that quotes were needed for a meaningful result); of these, 20
queries combined quotes and the asterisk, and 23 combined quotes with the site operator;
6 instances were noted in which all three operators were combined in one query.
In general, this translator spent a considerable amount of time and effort on her
language research, often trying different operators or replacing some components of
queries with possible alternatives when verifying a single item. This tendency is apparent
from the following examples:
Ex. 1. When comparing two different collocations, “take a pill” vs. “swallow a pill”,
the translator used not only a combination of the quotes and site operators but she later
also changed the infinitive form of both verbs to an –ing form in order to see the items
searched in more than one syntactic function.
Ex. 2. In order to verify her hypothesis that the preposition normally accompanying
the verb increase is by, the translator created a complex query, combining all three
operators: “increase * 10% in the last” site:uk. She then tried several modifications:
changing the verb form to increased, omitting the figure 10 and expanding the site
operator to site:bbc.co.uk. And although she found that sometimes the preposition is
omitted (failing to notice that this is normally only done in newspaper headlines), she
preferred to use by found in the majority of the hits.
Ex. 3. In an attempt to find an idiomatic formulation concerning the US being the top
consumer of dietary supplements, the translator tried multiple variations by changing
the name of the country and the numeral, and by deleting and adding words: moving
from “the US * the first place in” to “the UK * the first place in using” and “the US
take the * place in”, she later found that the query “first place worldwide” returned a
large number of hits; she inspected the results carefully, noticing that a definite article
and two prepositions in are needed for the phrase to be usable in a sentence.
Nevertheless, two examples illustrate that good search techniques do not guarantee
success on their own. In Ex. 4, the comparison of “a number of people have” vs. “a
number of people has”, using the quotes and, later, site:uk, led the user to choose the
former as it appears in many more results than the latter. And although the phrase is
correct in terms of grammar, it is wrong in the context at hand as it carries a completely
different sense than the correct the number of people, which requires the singular has.
Ex. 5. In another attempt to confirm her intuition, she compared a phrase she considered
correct, “too many food supplements”, with the much less acceptable “too much food
supplements”. Finding that the latter had many more hits, she opted for it, although
– as she pointed out in the protocol – this was contrary to her belief. In this instance, the
student failed to exercise her usual critical judgment, for it is evident on closer inspection
that many of the results returned for “too much food supplements” reflect a different
linguistic context to that investigated by the translator. Typically, one part of the query
116 D. Mraček
belongs to a different sentence than the rest of the query; it is apparent that much is a
modifier of the uncountable food not the countable supplements, cf. “So, do not eat too
much food. Supplements also provide…”.
Finally, two examples are worth a closer look as they illustrate the user’s meticulous
approach to the text production task at hand, the object of verification being punctuation.
In Ex. 6, the translator compared “In the United States, a number of” vs. “In the United
States a number of”. With the GSE being punctuation insensitive, it makes no difference
whether a comma is used or not. Still, the user was able to notice in the examples
returned that when placed initially, adverbials such as in the United States are separated
by a comma from the rest of the sentence. Similarly, a quick glance at the results returned
for the queries “a number of people, who” and “a number of people who” enabled her
to opt for the correct version without a comma in Ex. 7.
In sum, participant No. 2, who attended both GSE demonstrations, did systematic
and well-considered research, using multiple queries which involved whole phrases and
clauses (and even paying attention to punctuation, an issue ignored by the other three
translators), often combining different operators and manifesting no small degree of
creativity in a laudable effort to gain access to authentic linguistic material. Unlike
participant No. 1, she was consistently rigorous throughout the exercise, although some
of the queries led her to revise what had originally been a correct formulation.
On the other hand, a few minor infelicities could have been corrected had the transla-
tor made consistent use of the GSE. For instance, the use of articles in fixed phrases can
be verified efficiently via the GSE: the search “multivitamins on the daily basis” (Ex.
1) returns no hits compared to almost 2,000 hits for “multivitamins on a daily basis”.
Similar results are achieved with other words which can reasonably be expected in the
syntactic position before the fixed phrase, e.g. “taken on a daily basis” vs. “taken on the
daily basis”. Therefore, the key idea that learners need to internalize is that while queries
need to be long enough to return authentic linguistic material, individual words which
are less directly relevant to the issue being verified can be replaced with items that are
more likely to occur in that position. The phrase “on a prescription” (Ex. 2, where “on
prescription” is correct) is another example of a mistake avoidable through rigorous use
of the GSE.
In the few queries made, the translator sometimes reported relying on, or being
assisted by, her language instinct:
Ex. 3. She first made a total of 5 queries with the asterisk operator to explore the
possible verbs linking the words medicine, minerals, supplements and human body (e.g.
“medicine * in human body”, “supplements * in a human body”). Having found no
solution acceptable to her, she tried substituting the asterisk with two verbs she thought
might be used. The comparison of “minerals congest in human body”, “minerals pile up
in human body”, “medicine piles up in human body” and “medicine piled up in human
body” brought no convincing results and so the translator followed her intuition (while
adding, correctly, the definite article), producing a relatively acceptable version.
Ex. 4. By comparing “experts agree on one thing” vs. “experts agree on one point” she
came to the conclusion that both versions are acceptable but the former is used more
frequently – and sounds better to her – than the latter.
Ex. 5. The translator decided to verify the phrase “experience of doctors show”, using
the site:uk operator to narrow the results down to British websites. She found no results,
failing to realize that this was caused by her mistake in subject-verb concord (singular
noun vs. plural verb). She then modified the query to “experiences of many doctors
show” and, using the quotes operator only, found no results again; this is probably to
do with the fact that unlike its direct Czech equivalent, the English word experience is
normally used in the singular when referring to the collection of knowledge acquired
over a long period of time. The translator then shortened the query to “experiences
of many doctors”; the GSE returned 9 instances, which the translator believed to be
a satisfactory result. She then finalized the segment by using her language knowledge
and/or intuition, adding, correctly, the definite article and a new verb, illustrate, although
her initial choice, show, would have been a much more natural collocate.
Ex. 6. After finding very few occurrences of the collocation “overconsumption of
medicine”, the user shifted her focus onto the word overconsumption on its own, first
using the quotes only, later adding site:uk. That brought her to what she calls “a sat-
isfactory number of results (including the Guardian)”. On the one hand, her ability to
assess the reliability of the source offered (based on the renown of the British daily)
is commendable; on the other hand, had she attempted different modifications of col-
locations involving the word, e.g. “the overconsumption of dietary supplements”, her
solution could have been informed more by hard data than her instinct.
118 D. Mraček
In sum, participant No. 3, who attended only one GSE session, used the asterisk
and site operators in addition to the basic quotes operator, demonstrating, in more than
one instance, a willingness to modify the query for more convincing results. It can be
argued, however, that a more systematic use of the GSE, based on less intuition and
more reflection on the possibilities of grammar, as well as a generally higher number of
queries and segments verified would have resulted in a more acceptable English text.
in the sentence would yield more convincing results again, e.g. “in the last 20 years”
site:uk vs. “in last 20 years” site:uk. Alternatively, the word years can be replaced with
months or weeks and the site operator can be made more specific (e.g. site:bbc.co.uk).
As they were, the queries were so general that the user was left to decide solely based on
the total number of hits shown for each, instead of considering at least some contextual
information.
The fact that the immediate context of the item searched is crucial (perhaps even
more so than the overall number of results) can be illustrated with Ex. 4. The user verified
the phrase “number of people”, using the quotes operator only. Seeing a huge number
of results (including in an Oxford dictionary, a trustworthy source), he used the phrase
without thinking too much about grammar rules. Had he paid more attention to how the
item was used in entire sentences (looking carefully at the first two or three pages of the
results returned by the GSE would have sufficed), he would probably have noticed that
(1) the definite article is a constant component of the phrase and (2) the -who-sentence
which follows the phrase is normally not preceded by a comma, being a defining clause.
The three wrongly placed commas before that are the result of the same interference of
his native language, which, too, could have been avoided if careful reading of the results
had been part of the search strategy.
On the other hand, in at least two cases, the translator searched entire phrases. For
instance, go down the same path was checked for naturalness, although an incorrect
verb form was used (Ex. 5; are going would be more precise than will). As for can even
lead to death (Ex. 6), the phrase was compared with can end up with death and the
former was chosen on the basis of frequency. In one instance (Ex. 7), a word offered by
a dictionary, overuse, was first searched with the quotes only, then site:uk was added,
and later the query was expanded to cover an entire collocation: “medication overuse”
site:uk, followed by “supplements overuse” site:uk and “supplement overuse” site:uk;
as the latter two yielded very few results, the former query was repeated and the user,
seeing enough hits, decided that overuse can indeed be used in the context of medicine.
In sum, participant No. 4, who attended only one GSE session, did no small number
of searches but they were rarely strategic: either they contained too few words or were
deemed successful even though the immediate context was not taken into consideration.
The aim of the present paper was to shed more light on the role that the Google Search
Engine can play in producing better-quality language, be it by a foreign-language learner
or a translation trainee, with students assuming both these roles in some educational
contexts. Rather than attempting a detailed quality assessment, the analysis focused on
the different strategies that learners adopt when confronted with the realities of, on the
one hand, L2 text production and, on the other, an online search engine.
The analysis of the data gleaned from the respondents’ translations, search histories
and verbal protocols revealed, first and foremost, a considerable amount of variation
among participants of the experiment as to the number of queries made in relation to
the same task, which consisted in the translation of a short Czech text into English, the
participants’ L2. The lowest (13) and highest (83) number of queries was made by the
120 D. Mraček
two respondents who attended two sessions dedicated to GSE, while those who only
attended one made an average of 30 searches. The use of combinations of two operators
(quotes plus site, quotes plus asterisk) was comparable in both groups, although only
one participant (having attended both sessions) was prepared to make a regular use of a
combination of all three operators. In general, queries served to verify frequent issues
pertaining to English grammar and vocabulary, with articles, pronouns, prepositions and
collocations leading the list. Occasionally, punctuation was the object of verification
and in some instances, the GSE provided access to background materials (such as online
dictionaries and articles on points of grammar).
The aspects of GSE research in which the two pairs were found to differ most
were the length and structure of queries, where those students who had attended two
sessions made longer queries which comprised entire complex phrases and clauses rather
than single words and simple phrases. Moreover, one of the more advanced students
demonstrated a willingness to invest time and cognitive effort to the search process,
making numerous modifications to her queries in order to explore as best as she could
the way in which the item is used in authentic language. Without making strong claims
about the representativeness of the sample, it could be argued that every additional
demonstration or controlled practice session may enhance the learner’s command of GSE
search strategies and, even more importantly, her understanding of how these strategies
need to complement her use of existing and developing language knowledge.
Indeed, it appears reasonable to suppose that the need to verify items of language
depends to a large extent on the learner’s or translator’s current level of target language
skills, and there was undoubtedly a degree of heterogeneity among the participants with
respect to language skills. No less importantly, the level of language knowledge impacts,
in turn, on the learner’s online search dexterity. This leads Fujii to say that the GSE “has
strong potential for advanced learners, who are resourceful enough to generate a higher
number of syntactic/lexical combinations” [3]. With less advanced learners of a foreign
language, a text production exercise involving the use of a search engine would be
considerably more time-consuming.
However, another factor may be at play here, one that has less to do with language
knowledge. As the analysis made obvious, the four participants differed with regard to
their personalities, some exhibiting, to a larger degree than others, reflection, healthy
skepticism, attention to detail and other qualities conducive to successful research.
While the study only relied on a modest sample of four participants, it has pointed
to the potential that the GSE has for improving the quality of a translator’s or learner’s
written output in L2. As they work on their text production task, language users (including
native speakers of that language) can complement their language skills and intuition with
systematic search procedures involving simple words and, more effectively, complex
lexical and syntactic structures, thereby increasing their control of the language material
they produce.
On a more general educational level, the integration of search engines into the learn-
ing process is in line with transformationist approaches as such activities work with
what the student already knows (grammar rules and the use of search engines in real
life situations), transforming this knowledge into higher-level skills where language and
technology combine to assist in the production of better-quality L2 communication.
Uncovering the Potential of the Google Search Engine 121
The teacher here acts in her capacity of facilitator, pointing students to more effective
strategies for online search and, more generally, language learning.
At the same time, the GSE can contribute to learner autonomy: when provided the
basic know-how by the teacher (during a face-to-face tutorial where a demonstration of
search tips is given together with instructions for home practice), learners can uncover
the potential of the tool for their own learning process, exploring for themselves and self-
regulating on the basis of individualized feedback as long as it includes an assessment
of the quality of the search process in addition to the quality of language.
At the methodological level, it appears that gaining a meaningful insight into the
thought and search processes can only be made possible by triangulation. As Ericsson
and Simon [4, cited in 5] stressed, think-aloud data from working memory will always
be incomplete and exclude a number of thought processes which are not held in working
memory long enough to be expressed verbally. It has, therefore, become a common
practice to supplement think-aloud data with other data-collection strategies. In the
present study, the use of search history turned out to be effective in providing more
details where verbalizations were incomplete or even inaccurate in the protocols.
Future investigations into the use of search engines in L2 translation and FL learn-
ing may attempt to establish a firmer link between the quality of language output, the
student’s level of language skills and online search skills. To that end, a more robust and
interdisciplinary research framework may be needed, possibly involving different data
collection methods such as observation, video recording, and interviews.
References
1. Nasser, L.: Exploring the benefits of Google trends. The Next Ad, online (2018). https://www.
thenextad.com/blog/exploring-the-benefits-of-google-trends/. Accessed 23 June 2019
2. Mraček, D.: The Google search engine: a blended-learning tool for student empowerment.
In: Proceedings of the International Symposium on Educational Technology (ISET 2019),
pp. 224–229. IEEE, Piscataway (2019). https://doi.org/10.1109/iset.2019.00054
3. Fujii, Y.: Making the most of search engines for Japanese to English translation: benefits and
challenges. Asian EFL J. 23, 41–77 (2007). http://www.asian-efl-journal.com/pta_Oct_07_yf.
pdf. Accessed 12 June 2019
4. Ericsson, K.A., Simon, H.A.: Verbal reports as data. Psychol. Rev. 87(3), 215–251 (1980)
5. Charters, E.: The use of think-aloud methods in qualitative research: an introduction to think-
aloud methods. Brock Educ. 12(2), 68–82 (2003)
The Effect of Gamification in User Satisfaction,
the Approval Rate and Academic Performance
danielsalas@correo.unicordoba.edu.co
1 Introduction
In Ecuador, the numbers of allegations of sexual abuse are dispersed and incomplete but
as well they are worrying. The data of the Office of Criminal Policy of the Prosecutor’s
Office, between 2014 and 2017 indicated the existence of 13,671 complaints of sexual
abuse [11]. In this sense, UNICEF and the Ecuador Government are carrying out different
campaign such as “Open your Eyes”, with the purpose of informing and sensitizing
families and society about this critical problem.
The CEC-EPN aware of this situation, consistent with the UNICEF Ecuador cam-
paign, and in cooperation with several inter-institutional organizations such as the Asso-
ciation for Children and the Family, intend to support the abuse prevention through
the generation of free training strategies, among them Massive Online Open Courses
(MOOC). However, it was identified the courses offered by the CEC-EPN suffers com-
mon problems related to low student participation, low academic performance, low
approval rate and low student satisfaction, being necessary to explore strategies to
improve the participants experience in the offered courses.
Currently, the concept of Gamification is being used in different contexts: business,
education, health, government and even in daily tasks [1]. This is due to the benefits of
applying different elements from games in non-play contexts, especially considering the
influence it has on people’s behavior, through stimulating motivation.
According with Rodriguez and Santiago: “Gamification is a process by which
mechanics and game design techniques are applied, to seduce and motivate the audience
in the granting of certain objectives” [2].
In educational settings, Gamification has reported benefits in the learning process.
Mateo [6] in his thesis concludes, at a theoretical level, that gamification brings important
advantages to the dynamization of the work climate through motivating and awakening
the interest of the student. Sánchez [7] describes the implementation of a mobile applica-
tion implementing gamification in a secondary school with the aim of achieving greater
motivation and adherence to healthy oral habits. Their results showed that 100% of the
students in the control group showed interest in the gamified strategy and 47% consider
that educational gamification is more interesting than the master class. The gamification
strategy, the contents, the procedures and activities were widely accepted. There was
also an increase of 27% of students who improved their oral health habits. On the other
hand, the mixed study conducted by Vélez [8, 12] carried out through an experimental
design demonstrated that the students of the gamified group showed higher attention
spans than the students of the master class group. In addition, the students of the gam-
ified group showed greater motivation to participate in the class and greater interaction
both in student-student and teacher-student relationships. This result supports the thesis
of the present project in which it is stated that a gamification strategy affects the level
of user interaction. In the same line, the studies of Jorge-Soteras [9] or INSERVER [9]
also show gamification can improve the results obtained by the students in educational
processes.
On the base of the literature, the present the study describes the use of a gamification
strategy based on role-play, applied to the MOOC of Preventive Strategies for Child
Sexual Abuse offered by the CEC-EPN, aimed at parents, family and teachers, as a
contribution to support the awareness of the Ecuadorian population on the issue of
prevention of child sexual abuse. The validations was oriented to measure the measures
user satisfaction, the approval rate and academic performance of the participants to
compare the participants’ behaviors in both, a traditional course and a gamified course.
The paper is divided into fourth sections. The first section correspond to the introduc-
tion that introduce the problem and supported literature. The second section describes
the design of the gamification strategy. The third section describes the methodology
124 G. Martínez et al.
applied in order to assess the level of impact of the gamification strategy, and finally, the
fourth section presents the conclusions obtained and the future work that the obtained
results could lead to.
Several methodologies and models have been defined to support the design of a gami-
fied system [2, 3, 5]. In the present study we have used the proposal of formal design
frameworks by Marczewski [2–4, 10] because it fix the requirements of our context and
strategy. The model consists of two loops:
(1) First Loop, Definition and Design. The following elements are defined:
(2) Second Loop, Construction of the Solution. The second loop is designing the user’s
journey, which will be the basis of the mechanics, motivations and behaviors that
will be used in the gamification. The user’s journey consists of four stages:
• Discovery: It is the mechanism used for the user in order to discover how to
perform the gamification strategy.
• On Board: It is the tutorial that allows the user to understand how the game will
develop.
• Immersion: It is the longest phase in which the user interacts with the mechanics
of available games.
• Master: It is the stage in which the user has reached the maximum level.
• Repetition: It is the stage in which it is decided if it makes sense for the user to
play again.
As mentioned the main purpose of the gamification strategy designed in this study was
support the prevention of child sexual abuse, but also improve the participants experience
in the course offered by the CEC-EPN.
The definition of the type of user was made through an initial survey that allows
defining the general characteristics of the user, and in this way identify the type of
player using Marczewski’s classification [4] defining 6 types of players: Disruptors,
Free Spirit (Explorers), Achievers, Players, Socializing and Philanthropists.
The users definition was carried out through the application of a short test similar
to that of Bartle Play Time Test1 to potential participants. The results show a greater
percentage of potential players of the Explorer type (32.3%) and Achiever (26.9%), as
shown in Fig. 1. This definition was useful for gamification mechanics.
Fig. 1. Results of the application of the test for player type analysis (Designed by the authors)
• The success of the gamification strategy that means to achieve improvement in user sat-
isfaction, approval rate and academic performance has been verified in the evaluation
of the strategy presented in this paper.
• The players’ success happens when the participants rescue children and families who
are at risk of abuse.
• The academic success is achieved when the participants successfully learn about
prevention strategies.
The repetition or “Replay” is placed in each level of game, when the user loses their
three lives, the game will be restarted.
• Discovery: when the user enters the gamified content, he is invited to assume the role
of Great Psychologist and complete the safety route by rescuing the children victims
of abuse. The condition to access the game will be to read the material corresponding
to the lessons. The material is in a downloadable format, so that it can be consulted
at any time.
• On Board: once the user confirms the beginning of the gamified content, the user
receives the challenge or conflict and the rules of the gamification. Each level has its
own challenges and rules. The user must express the understanding of them before
starting.
• Immersion: in the immersion phase, the “Great Psychologist” will enter the educa-
tional scenario in which the challenge will be carried out; in it he/she will find the
gamification mechanics associated with the different types of player according with
Table 1.
Table 1. Mechanics and dynamics according to the type of player (Designed by the authors)
Table 1. (continued)
In addition to these mechanics, some game elements were used in order to create a
fun environment. These elements are presented in Table 2.
Elements Description
Fantasy and Theme An educational scenario in which the main character, “Great
Psychologist” is hired by Ana, the director, to solve the challenges, was
designed. The narrative emerges as a dialogue between the two personals
within which the challenges of the game are silvered
Progress/Feedback A progress bar known as “Completion Status” is integrated into the
virtual platform, in such a way as to allow the user to verify the progress
he has within the course
(continued)
128 G. Martínez et al.
Table 2. (continued)
Elements Description
Aversion to loss A system of gain and loss of points and lives was designed. The loss of
points originates when the user makes an incorrect selection of the
presented options; When the point marker reaches 0 the user loses a life.
If you lose both lives the game restarts. If the user use one of the clues to
solve the challenge, he/she lose one point for each track used
Rules Although each of the challenges has its rules, the following basic rules
were defined:
• To select the correct answer, you must click on the option that you
believe is correct
• Each option has two tracks, by using the tracks you will subtract one
point per track
• The user starts the gamified strategy with two lives, each life gives him
10 points
• If you choose the correct option you will receive 10 points, if you
choose the wrong option you will lose 10 points
• Each time the score reaches 0 you will lose a life
• If he loses all lives he will have to start the gamified strategy again
• The user will go to the next level when he has successfully resolved the
challenge
Table 3. Blocks in the Moodle platform according to the Traditional and Gamified courses
Table 3. (continued)
3 Evaluation
3.1 Description
The control intervention was carried out in the context of a web-based traditional
course implemented under ADDIE model without gamification.
The experimental intervention was carried out in the context of the MOOC
implemented using the gamification strategy defined in Sect. 2.
At the end of the course an analysis of the results obtained was carried out doing
several comparisons: Comparison of the level of user satisfaction, Comparison of the
Approval Rate, Comparison of academic performance and Comparison of academic
performance.
3.2 Description
The results found in the Satisfaction of the User category show in Fig. 2, do not reflect
a significant level of impact of the strategy of gamification on satisfaction considering
both the control group and the experimental group are in the Fully Satisfied category,
with a minimum difference of percentages: 94.9% for the control group and 93.7% for
the experimental group.
The elements that make up user satisfaction are: the Methodology, Evaluation and
Administrative Technical Management. When evaluating these items, it is possible to
observe that users of the gamified course show greater satisfaction with the methodology
of the course (95.6%) in relation to the control group (95.6%); However, the difference
is not significant either.
When analyzing the approval rate results, the difference between the groups is sig-
nificant. As shown in Fig. 3, the control group obtained an approval rate of 42% while
the experimental group a rate of 56%. Therefore, considering this variable, the impact
of the gamification strategy on the approval of the course is certainly higher.
Regarding the academic performance, the users of the control group obtained an
average of 41.31 points out of 100 in performance, while the experimental group obtains
an average of 58.30. The comparison of the distribution of users by academic perfor-
mance reflects that there are more users of the experimental group at the high level
(Experimental Group 44% vs. Control Group 33%). At the same time, when the aver-
ages distributed by groups are compared, the ratings’ average of the experimental group
are higher than those of the control group in all cases, as shown in Fig. 4.
The Effect of Gamification in User Satisfaction 131
Based on the results presented by other authors and on the theoretical foundation of gam-
ified systems design, a strategy of gamification based on role play has been implemented
in a MOOC on Prevention of Child Sexual Abuse offered by CEC-EPN.
Our research shows for a particular case, the power of gamification in educational
contexts, offering positive findings about its benefits on academic performance and the
approval rate. One of the limitations of this study is that its application has been reduced
to the implementation of the strategy in a MOOC course with a specific topic. However,
starting with the results obtained, this can be replicated in other courses.
The strategy has been built applying a particular model of formal design, in this sense
it is limited to the elements and the procedure established by Marczewski. However, our
interest for future work is to apply other design models.
132 G. Martínez et al.
References
1. INSERVER: Gamification against bullying (2018)
2. Jorge-Soteras, M.: Application of gamification for the improvement of a didactic unit in higher
vocational training. International University of La Rioja (2017)
3. Mateo, C.: Application of the methodology of the gamification through TIC in 3° of ESO.
Universidad Internacional de la Rioja (2016)
4. Marczewski, A.C.: Even Ninja Monkeys Like to Play. CreateSpace Independent Publishing
Platform (2015)
5. Mora, A., Riera, D., Gonzalez, C., Arnedo-Moreno, J.: A literature review of gamification
design frameworks. In: 7th International Conference on Games and Virtual Worlds for Serious
Applications (VS-Games), pp. 1–8. Skovde (2015)
6. Park, S., Kim, S.: An optimized number of game mechanics and PLEX fun factors for the
gamification development. J. Digital Contents Soc. 19, 2009 (2018)
7. Rodriguez, F., Santiago, R.: Gamification: how to motivate your student and improve the
climate in the classroom. Editorial CEANO S.L.U, Barcelona (2015)
8. Sánchez, N.: Gamificación educativa a través de aplicaciones móviles en 1o ESO. Universidad
Internacional de la Rioja (2018)
9. The Marczewski framework. https://www.youtube.com/watch?v=506cWlmhghE&feature=
youtu.be. Accessed 01 Jul 2019
10. Tondello, G.F., Wehbe, R.R., Diamond, L., Busch, M., Marczewski, A., Nacke, L.E.: The
gamification user types hexad scale. In: Proceedings of the 2016 Annual Symposium on
Computer-Human Interaction in Play, pp. 229–243. New York, NY, USA (2016)
11. Vanégas, A: Cifras y denuncias sustentan la pregunta en la consulta sobre abuso
de niños. EL UNIVERSO. https://www.eluniverso.com/noticias/2017/10/24/nota/6447337/
cifras-denuncias-sustentan-pregunta-sobre-abuso-ninos. Accessed 01 Jul 2019
12. Vélez, I.: The gamification in the learning of the university students. Traces Faces 18, 1–12
(2016)
Topic Detection for Online Course
Feedback Using LDA
1 Introduction
Online learning has been on the rise in recent years. Students can fit them around
their existing responsibilities and commitments, and can engage with multimedia
content and learning materials at whatever time is most convenient to them.
Online courses are easily accessible on much smaller budgets. In addition to
the convenience and the cost, a large number of students are turning to online
learning courses because they have become a better way to learn [30,34]. Those
students who are serious about improving their understanding, learning new
skills and gaining valuable qualifications are keen to enroll in the type of course
that will be the most effective.
In an online courses, students may deliver various feedback to evaluate every-
thing they experience online. They share learning feelings about online courses,
which provides many opportunities to discover students’ emotional states. How-
ever, the unstructured textual data may pose a difficulty for teachers who want
to understand the feedback. Therefore a system to summarize all student feed-
back and giving an overall summary will be very useful for teachers, lecturers,
schools, universities, and all education systems to attend to the student feedback
and to improve the education system [5,19,23].
With the large amount of feedback, it is difficult for people to comprehend
a large number of comments in a chronological order and monitor student com-
ments manually. Several approaches are proposed to summary student course
c Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 133–142, 2020.
https://doi.org/10.1007/978-3-030-38778-5_16
134 S. Unankard and W. Nadee
2 Related Works
Education research has studied in various areas in order to improve the design
and delivery of the academic programme. Student course feedback provides
Topic Detection for Online Course Feedback Using LDA 135
3 Proposed Approach
3.1 Pre-processing
In order to improve the quality of our dataset and the performance of the sub-
sequent steps, the pre-processing was designed to ignore common words that
carry less important meaning than keywords and remove irrelevant data. The
comments are converted into lower case and are removed by a single charac-
ter word. The stop words are removed and all words are converted into a seed
word (stemming word) by using Lucene 3.1.0 Java API1 . All comments after
pre-processing are stored in the database.
1
http://lucene.apache.org.
Topic Detection for Online Course Feedback Using LDA 137
Variable Description
M Number of comments
N Number of unique words (vocabulary)
α Dirichlet prior for Θ
β Dirichlet prior for Φ
Θm Topic distribution in comments
Φk Word distribution in topics
z Topic Assignment of Words
w Words in documents
2
https://www.coursera.org/.
3
https://www.kaggle.com/septa97/100k-courseras-course-reviews-dataset.
Topic Detection for Online Course Feedback Using LDA 139
Table 2. (continued)
However, there are some limitations for LDA. The number of topics is fixed
and must be known ahead of time and no evolution of topics over time. Despite its
limitations, LDA is still central to topic modeling and has really revolutionized
the field.
5 Conclusions
In this paper, we proposed an approach for topic detection of online course feed-
back by utilizing Latent Dirichlet Allocation (LDA). We describes our demon-
stration system for detecting feedback topics over online course system. Our
system is able to summarize topics and present to instructors via Wordcloud.
The course feedback related to Machine Learning course from the website of
Coursera is used to demonstrate the effectiveness of our approach. For future
work, the further performance evaluation will be performed and opinion anal-
ysis for student course feedback will be studied to discover students’ emotions.
Moreover, student learning behavior patterns will be studied.
References
1. Allahyari, M., et al.: Text summarization techniques: a brief survey. arXiv preprint
arXiv:1707.02268 (2017)
2. Alvanaki, F., Michel, S., Ramamritham, K., Weikum, G.: See what’s enblogue:
real-time emergent topic identification in social media. In: Proceedings of 15th
International Conference on Extending Database Technology, EDBT 2012, Berlin,
Germany, 27–30 March 2012, pp. 336–347 (2012)
3. Blei, D.M.: Probabilistic topic models. Commun. ACM 55(4), 77–84 (2012).
https://doi.org/10.1145/2133806.2133826
4. Blei, D.M., Ng, A.Y., Jordan, M.I.: Latent dirichlet allocation. J. Mach. Learn.
Res. 3(1), 993–1022 (2003)
5. Bonnel, W.: Improving feedback to students in online courses. Nurs. Educ. Per-
spect. 29(5), 290–294 (2008)
6. Cataldi, M., Caro, L.D., Schifanella, C.: Emerging topic detection on twitter based
on temporal and social terms evaluation. In: Proceedings of the Tenth International
Workshop on Multimedia Data Mining, pp. 4:1–4:10 (2010)
7. Chathuranga, J., Ediriweera, S., Hasantha, R., Munasinghe, P., Ranathunga, S.:
Annotating opinions and opinion targets in student course feedback. In: Proceed-
ings of the Eleventh International Conference on Language Resources and Evalu-
ation (LREC-2018) (2018)
8. Cohen, R., Ruths, D.: Classifying political orientation on twitter: it’s not easy!.
In: Proceedings of the 7th International Conference on Weblogs and Social Media,
ICWSM 2013, pp. 91–99, January 2013
9. Cristani, M., Perina, A., Castellani, U., Murino, V.: Geo-located image analysis
using latent representations. In: 2008 IEEE Conference on Computer Vision and
Pattern Recognition, pp. 1–8, June 2008
10. Eisenstein, J., O’Connor, B., Smith, N.A., Xing, E.P.: A latent variable model for
geographic lexical variation. In: Proceedings of the 2010 Conference on Empir-
ical Methods in Natural Language Processing, pp. 1277–1287. EMNLP 2010.
Association for Computational Linguistics, Stroudsburg (2010). http://dl.acm.org/
citation.cfm?id=1870658.1870782
Topic Detection for Online Course Feedback Using LDA 141
11. Fan, X., Luo, W., Menekse, M., Litman, D., Wang, J.: Coursemirror: enhanc-
ing large classroom instructor-student interactions via mobile interfaces and nat-
ural language processing. In: Proceedings of the 33rd Annual ACM Conference
Extended Abstracts on Human Factors in Computing Systems, pp. 1473–1478.
ACM (2015)
12. Fang, Y., Si, L., Somasundaram, N., Yu, Z.: Mining contrastive opinions on polit-
ical texts using cross-perspective topic model. In: Proceedings of the Fifth ACM
International Conference on Web Search and Data Mining, pp. 63–72. WSDM
2012. ACM, New York (2012). http://doi.acm.org/10.1145/2124295.2124306
13. Garcı́a-Hernández, R.A., Montiel, R., Ledeneva, Y., Rendón, E., Gelbukh, A.,
Cruz, R.: Text summarization by sentence extraction using unsupervised learning.
In: Gelbukh, A., Morales, E.F. (eds.) MICAI 2008. LNCS (LNAI), vol. 5317, pp.
133–143. Springer, Heidelberg (2008). https://doi.org/10.1007/978-3-540-88636-
5 12
14. Goorha, S., Ungar, L.H.: Discovery of significant emerging trends. In: Proceedings
of the 16th ACM SIGKDD International Conference on Knowledge Discovery and
Data Mining, Washington, DC, USA, 25–28 July 2010, pp. 57–64 (2010)
15. Hasan, K.S., Ng, V.: Automatic keyphrase extraction: a survey of the state of the
art. In: Proceedings of the 52nd Annual Meeting of the Association for Computa-
tional Linguistics, vol. 1: Long Papers. pp. 1262–1273 (2014)
16. Hong, L., Davison, B.D.: Empirical study of topic modeling in Twitter. In: Pro-
ceedings of the First Workshop on Social Media Analytics. SOMA 2010, pp. 80–88.
ACM, New York (2010). http://doi.acm.org/10.1145/1964858.1964870
17. Paul, M.J., Dredze, M.: You are what your tweet: analyzing twitter for public
health. Artif. Intell. 38, 265–272 (2011)
18. Linstead, E., Rigor, P., Bajracharya, S., Lopes, C., Baldi, P.: Mining concepts
from code with probabilistic topic models. In: Proceedings of the Twenty-second
IEEE/ACM International Conference on Automated Software Engineering. ASE
2007, pp. 461–464. ACM, New York (2007). http://doi.acm.org/10.1145/1321631.
1321709
19. Lumpkin, A., Achen, R.M., Dodd, R.K.: Student perceptions of active learning.
Coll. Stud. J. 49(1), 121–133 (2015)
20. Luo, W., Fan, X., Menekse, M., Wang, J., Litman, D.: Enhancing instructor-
student and student-student interactions with mobile interfaces and summariza-
tion. In: Proceedings of the 2015 Conference of the North American Chapter of
the Association for Computational Linguistics: Demonstrations, pp. 16–20 (2015)
21. Luo, W., Litman, D.: Summarizing student responses to reflection prompts. In:
Proceedings of the 2015 Conference on Empirical Methods in Natural Language
Processing, pp. 1955–1960 (2015)
22. Luo, W., Liu, F., Litman, D.: An improved phrase-based approach to annotating
and summarizing student course responses. arXiv preprint arXiv:1805.10396 (2018)
23. Luo, W., Liu, F., Liu, Z., Litman, D.: Automatic summarization of student course
feedback. arXiv preprint arXiv:1805.10395 (2018)
24. Martins, A.F., Smith, N.A.: Summarization with a joint model for sentence extrac-
tion and compression. In: Proceedings of the Workshop on Integer Linear Program-
ming for Natural Language Processing, pp. 1–9. Association for Computational
Linguistics (2009)
25. Mathioudakis, M., Koudas, N.: Twittermonitor: trend detection over the twitter
stream. In: Proceedings of the ACM SIGMOD International Conference on Man-
agement of Data, SIGMOD 2010, Indianapolis, Indiana, USA, 6–10 June 2010, pp.
1155–1158 (2010)
142 S. Unankard and W. Nadee
26. Mehrotra, R., Sanner, S., Buntine, W., Xie, L.: Improving LDA topic models for
microblogs via tweet pooling and automatic labeling. In: Proceedings of the 36th
International ACM SIGIR Conference on Research and Development in Informa-
tion Retrieval, pp. 889–892. ACM (2013)
27. Ramage, D., Dumais, S., Liebling, D.: Characterizing microblogs with topic models.
In: Fourth International AAAI Conference on Weblogs and Social Media (2010)
28. Rush, A.M., Chopra, S., Weston, J.: A neural attention model for abstractive
sentence summarization. arXiv preprint arXiv:1509.00685 (2015)
29. Schinas, M., Papadopoulos, S., Kompatsiaris, Y., Mitkas, P.A.: Visual event sum-
marization on social media using topic modelling and graph-based ranking algo-
rithms. In: Proceedings of the 5th ACM on International Conference on Multimedia
Retrieval, pp. 203–210. ACM (2015)
30. Shapiro, H.B., Lee, C.H., Roth, N.E.W., Li, K., Rundel, M.Ç., Canelas, D.A.:
Understanding the massive open online course (MOOC) student experience: an
examination of attitudes motivations and barriers. Comput. Educ. 110, 35–50
(2017)
31. Steyn, C., Davies, C., Sambo, A.: Eliciting student feedback for course devel-
opment: the application of a qualitative course evaluation tool among business
research students. Assess. Eval. High. Educ. 44(1), 11–24 (2019)
32. Sung, Y.T., Liao, C.N., Chang, T.H., Chen, C.L., Chang, K.E.: The effect of online
summary assessment and feedback system on the summary writing on 6th graders:
The LSA-based technique. Compu. Educ. 95, 1–18 (2016)
33. Thomas, S.W.: Mining software repositories using topic models. In: 2011 33rd
International Conference on Software Engineering (ICSE), pp. 1138–1139, May
2011
34. Toven-Lindsey, B., Rhoads, R.A., Lozano, J.B.: Virtually unlimited classrooms:
pedagogical practices in massive open online courses. Internet High. Educ. 24,
1–12 (2015)
35. Wan, X., Yang, J.: Multi-document summarization using cluster-based link anal-
ysis. In: Proceedings of the 31st Annual International ACM SIGIR Conference on
Research and Development in Information Retrieval, pp. 299–306. ACM (2008)
36. Wang, L., Ling, W.: Neural network-based abstract generation for opinions and
arguments. arXiv preprint arXiv:1606.02785 (2016)
37. Wang, Y., Agichtein, E., Benzi, M.: TM-LDA: efficient online modeling of latent
topic transitions in social media. In: Proceedings of the 18th ACM SIGKDD Inter-
national Conference on Knowledge Discovery and Data Mining, pp. 123–131. ACM
(2012)
38. Welch, C., Mihalcea, R.: Targeted sentiment to understand student comments. In:
Proceedings of COLING 2016, the 26th International Conference on Computa-
tional Linguistics: Technical Papers, pp. 2471–2481 (2016)
39. Xiong, W., Litman, D.: Empirical analysis of exploiting review helpfulness for
extractive summarization of online reviews. In: Proceedings of coling 2014, the
25th International Conference on Computational Linguistics: Technical Papers,
pp. 1985–1995 (2014)
40. Zhang, Y., Chen, M., Huang, D., Wu, D., Li, Y.: iDoctor: personalized and profes-
sionalized medical recommendations based on hybrid matrix factorization. Future
Gener. Comput. Syst. 66, 30–35 (2017)
SPeL (Social and Personal Computing
for Web-Supported Learning
Communities)
An Educational Model for Integrating
Game-Based and Problem-Based Learning
in Data-Driven Flipped Classrooms
1 Introduction
The past decades have seen a transformation in the way educators and course designers
approach education. The development of technology-enhanced learning appears as a
response to the needs of a new society and a major change in education [1]. Technolog-
ical developments herald a necessary change in learning paradigms and forms of tradi-
tional schooling. Conventional education focused mostly on transmission of knowledge.
Recent developments in educational technology focus on learning, i.e. the acquisition of
new mental schemes, knowledge, abilities, and skills that can be used to solve problems
successfully [2]. This resulted in an increased appreciation for active learning. Accord-
ing to Lee in [3], active learning “shifts the focus of learning from passively receiving
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 145–154, 2020.
https://doi.org/10.1007/978-3-030-38778-5_17
146 M. Algayres and E. Triantafyllou
2 Background
must study the initial problem formulation in order to formulate a concrete problem they
can solve. Then, they analyze the problem by gaining knowledge on the topics related
to it, and finally design, implement and evaluate a solution to it. Research has shown
that the flipped classroom combined with the problem-solving strategy was more effec-
tive than e-learning, or learning through isolated problem-solving sequences [11]. Lai
and Hwang similarly emphasized that a self-regulated flipped classroom approach can
improve students’ learning performance. Moreover, the problem-solving and personal-
ized interaction, which takes place face-to-face, sets these classes apart, making them
more effective than MOOCs [12].
Research into educational games is long established, with recent developments along
the fields of serious games [16], digital GBL [17], and gamification, the last defined
as “the use of video game elements to improve user experience and user engagement
in non-game services and applications” [18]. Games embody well-established prin-
ciples of active learning and allow replacing the learning activity into a meaningful
context [19].
Research in combining the FC and GBL is also a recent approach, which has gained
traction in recent years. Several studies highlighted how GBL and gamification could
improve the FC model, both for support in the pre-class process [20] or through gamified
activities in the classroom [21].
Furthermore, research into LA applied to GBL and gamification shows that there is
great potential in using these methodologies with EDM in the FC. For example, Klemke
et al. [7] proposed the model of the flipped MOOC to prepare the students for class,
using LA and gamification to track the students’ engagement with the learning material
during pre-class preparation. Other studies underline the fact that using these methods
in so-called smart classrooms could support adaptive learning, allowing students and
educators to tailor the learning process to individual needs [4]. For example, Yang [22]
developed the master learning theory to present a model supporting GBL and LA, where
gaming elements and personalized feedback reinforce the learning process in students.
Similarly, Busch et al. [23] presented the development of a learning mobile game model,
where LA were used in evaluating the success and fail rates of players with the aim to
support the learning experience.
The proposed educational model is based on the stipulation that learners should be able
to gain control and reflect on their learning process. Following the learning cycle in the
FC Model established by Gernstein [8], the learning sequence should allow students to
analyze, apply, and create based on knowledge that they have accessed, with opportu-
nities to understand and remember it. Finally, learners should be able to evaluate their
own progression to support their self-directed learning. With these objectives in mind,
we have devised a three-tier model that supports learning in FCs (Fig. 2).
The first layer of the model describes the learning activities that take place in the FC.
These activities are organized following the FC cycle of pre-class, in-class and post-class.
They are supported by the Learning Design methodology, whose successful application
in the FC has already been studied in [24]. Learning design “…is defined as the descrip-
tion of the teaching-learning process that takes place in a unit of learning (e.g., a course,
a lesson or any other designed learning event)” [25, p. 13]. The key principle in Learning
Design is that it represents “the learning activities and the support activities that are per-
formed by different persons (learners, teachers) in the context of a unit of learning” [25].
An Educational Model for Integrating Game-Based and Problem-Based Learning 149
The most common framework for Learning Design is defined in the Larnaca Declaration
on Learning Design [26]. This declaration defines the core concepts of Learning Design
(guidance, representation, sharing), and the development of a Learning Design activity
by three steps: planning activity, design core-learning concepts, and implement activity.
This approach allows embedding layers of PBL elements, gamification, and GBL
in the learning activity. While it would be impossible to provide an extensive list of the
learning activities on the design level, previous research provides us with examples of
combining such methodologies with the FC:
– Gamification: the use of gamification in the FC appears the most common, as elements
like badges and leaderboards are a common feature of Virtual Learning Environments
(VLEs) such as MOOCs. Klemke et al. [7] suggested the model of the flipped MOOC to
combine online preparation for class with gamification and Tsay et al. [27] similarly
tested a model of FC supported by gamification (with goals, variable difficulties,
badges and leaderboards), and tested it as more efficient than traditional learning in
terms of learning outcomes.
– GBL: educational games are flexible and diverse and appear both in pre-class prepa-
ration [28] and as an in-class activity. In-class games can help students to practice
notions [21], and they can also be used as simulations of real-life situations [29].
– PBL: research exploiting active use of PBL in the FC is still rare. Song et al. [30]
presented the FC as a means to exploit the classroom time for appropriately designed
interactive learning activities such as collaborative and PBL activities. Çakiroglu and
Özturk [31] carried PBL activities in FCs to promote self-regulation among under-
graduate students with positive results. They concluded that this method resulted in
150 M. Algayres and E. Triantafyllou
higher levels of goal setting, planning and task strategies, and promoted self-regulation
skills among the students.
In our framework, learning activities are inspired by the PBL approach and are
distributed along the FC cycle (Fig. 2). If the experiential engagement FC stage is
employed for introduction to the next FC cycle, learning activities aiming at introducing
students to an ill-defined problem can take place during this stage. Such activities could
include watching videos, listening to podcasts, or reading. During the concept exploration
FC stage, students can be invited to understand and analyse the ill-defined problem.
Problem analysis can be achieved by following video or audio tutorials, reading material
on websites or textbooks, or consulting experts, among other activities. For the meaning-
making FC stage, students are called to formulate a specific problem, which stems from
the ill-defined one, and that they will solve during the next stages of the FC. During this
process, students may discuss or brainstorm with their peers or their teachers through
discussion boards or blogs, and they may produce written material for arguing on the
selection of the problem. They may also get feedback either from peers or from teachers
on the problem formulation. Moreover, students may be invited to reflect on the problem
formulation or they may be assessed on the knowledge they gained so far, in order
to assure that they come to the in-class session prepared. Learning activities such as
audio-visual reflections, reflective videos/podcasts, and tests/quizzes can be employed
for this matter. For the demonstration and application FC stage, the students can work
in class on designing and implementing a solution to the selected problem. During this
stage, students can work in groups, while the teachers will support students on problem-
solving. Finally, during experiential engagement (when implemented as the final stage of
the FC), students can evaluate and reflect on the implemented solution by running out-of-
class experiments, surveys, or tests/simulations and producing reports on the evaluation
results. The aforementioned PBL/FC cycle does not presupposes that students have to
work with different problems in every cycle. Students may select a problem that will be
solved in several iterations. In every iteration, students may solve parts of the problem
or may re-define the problem.
In our framework, learning activities will also take opportunities to exploit GBL and
gamification to support the educational model. The whole of the FC is supported by a gam-
ification system that uses common features such as badges, levels and leaderboards to help
students track their own progress, visualize their progress and manage their own learn-
ing paths. Furthermore, GBL can be used at each step of the educational model to support
engagement in the learning activity. During the concept exploration process, quizzes and
digital GBL can be used to check that students have engaged with the learning material. In
the meaning-making stage, specific serious games can be used for students to practice the
new concepts that they have learned. In the demonstration and application stage, both gam-
ification and GBL can be implemented, especially during in-class sessions. Levels, badges
and leaderboards can be used to monitor students’ progression and help them visualise
the different steps of the learning process. Similarly, various forms of serious games can
be used as tools for application. In the last stage, experiential engagement, the debriefing
process of a serious game can be used as a means to engage the students in self-reflection
regarding the learning process and what they learned.
An Educational Model for Integrating Game-Based and Problem-Based Learning 151
4 Conclusion
In conclusion, our educational model aims at providing a new expanded model based on
the FC methodology, which can improve and get beyond the standard, and sometimes
152 M. Algayres and E. Triantafyllou
Acknowledgement. This research was conducted in the context of the FLIP2G project. This
project has been funded with the support of the Erasmus+programme of the European Union.
This paper reflects the views only of the authors, and the Commission cannot be held responsible
for any use which may be made of the information contained therein.
References
1. Chen, N.S., Cheng, I.L., Chew, S.W.: Evolution is not enough: revolutionizing current learning
environments to smart learning environments. Int. J. Artif. Intell. Educ. 26(2), 561–581 (2016)
2. Alonso, F., López, G., Manrique, D., Viñes, J.M.: An instructional model for web-based
e-learning education with a blended learning process approach. Br. J. Edu. Technol. 36(2),
217–235 (2005)
3. Frey, B.: The SAGE Encyclopaedia of Educational Research, Measurement, and Evaluation,
pp. 39–40. SAGE Publications, Thousand Oaks (2018)
4. Uskov, V.L., Bakken, J.P., Penumatsa, A., Heinemann, C., Rachakonda, R.: Smart pedagogy
for smart universities. In: Uskov, V.L., Howlett, R.J., Jain, L.C. (eds.) SEEL 2017. SIST, vol.
75, pp. 3–16. Springer, Cham (2018). https://doi.org/10.1007/978-3-319-59451-4_1
5. Abeysekera, L., Dawson, P.: Motivation and cognitive load in the flipped classroom: definition,
rationale and a call for research. High. Educ. Res. Dev. 34(1), 1–14 (2015)
6. Nechodomu, T., Falldin, M., Hoover, S.: CEHD Flipped Learning Guide (2016)
7. Klemke, R., Eradze, M., Antonaci, A.: The flipped MOOC: using gamification and learning
analytics in MOOC design—a conceptual approach. Educ. Sci. 8(1), 25 (2018)
8. Gerstein, J.: The flipped classroom: a full picture, User Generated Education (2011). http://
usergeneratededucation.wordpress.com/2011/06/13/the-flipped-classroom-model-afull-
picture/
9. Neville, A.J.: Problem-based learning and medical education forty years on. Med. Principles
Pract. 18(1), 1–9 (2009)
10. Barge. S.: Principles of Problem and Project Learning, the Aalborg PBL Model. Aalborg Uni-
versity, Aalborg (2010). http://www.aau.dk/digitalAssets/62/62747_pbl_aalborg_modellen.
pdf
11. Chiang, T.H.C.: Analysis of learning behavior in a flipped programing classroom adopting
problem-solving strategies. Interact. Learn. Environ. 25(2), 189–202 (2017)
12. Lai, C.L., Hwang, G.J.: A self-regulated flipped classroom approach to improving students’
learning performance in a mathematics course. Comput. Educ. 100, 126–140 (2016)
13. Long, P., Siemens, G.: Penetrating the fog: analytics in learning and education. Educause Rev.
Online 46(5), 31–40 (2011)
14. Lu, O.H., Huang, J.C., Huang, A.Y., Yang, S.J.: Applying learning analytics for improving stu-
dents’ engagement and learning outcomes in an MOOCs enabled collaborative programming
course. Interact. Learn. Environ. 25(2), 220–234 (2017)
An Educational Model for Integrating Game-Based and Problem-Based Learning 153
15. Doko, E., Bexheti, L.A.: A systematic mapping study of educational technologies based on
educational data mining and learning analytics. In: 2018 7th Mediterranean Conference on
Embedded Computing (MECO), pp. 1–4. IEEE (2018)
16. Michael, D.R., Chen, S.L.: Serious Games: Games That Educate, Train, and Inform. Muska
& Lipman/Premier-Trade, Boston (2005). Thomson Course Technology
17. Prensky, M.: Digital game-based learning. Comput. Entertainment (CIE) 1(1), 21 (2003)
18. Deterding, S., Sicart, M., Nacke, L., O’Hara, K., Dixon, D.: Gamification. using game-design
elements in non-gaming contexts. In: CHI 2011 Extended Abstracts on Human Factors in
Computing Systems, pp. 2425–2428. ACM (2011)
19. Van Eck, R.: Digital game-based learning: it’s not just the digital natives who are restless.
Educause Rev. 41(2), 16 (2006)
20. Jo, J., Jun, H., Lim, H.: A comparative study on gamification of the flipped classroom in
engineering education to enhance the effects of learning. Comput. Appl. Eng. Educ. 26(5),
1626–1640 (2018)
21. Hung, H.T.: Gamifying the flipped classroom using game-based learning materials. ELT J.
72(3), 296–308 (2018)
22. Yang, K.H.: Learning behavior and achievement analysis of a digital game-based learning
approach integrating mastery learning theory and different feedback models. Interact. Learn.
Environ. 25(2), 235–248 (2017)
23. Busch, C., Claßnitz, S., Selmanagic, A., Steinicke, M.: Developing and testing a mobile
learning games framework. Electron. J. e-Learning 13(3), 151–166 (2015)
24. Triantafyllou, E., Kofoed, L.B., Purwins, H., Timcenko, O.: Applying a learning design
methodology in the flipped classroom approach–empowering teachers to reflect and design
for learning. Tidsskriftet Læring og Medier (LOM), 9(15) (2016)
25. Koper, R.: Current research in learning design. J. Educ. Technol. Soc. 9(1), 13–22 (2006)
26. Dalziel, J., et al.: The Larnaca declaration on learning design (2013). http://www.
larnacadeclaration.org/
27. Tsay, C.H.H., Kofinas, A., Luo, J.: Enhancing student learning experience with technology-
mediated gamification: an empirical study. Comput. Educ. 121, 1–17 (2018)
28. Ling, L.T.Y.: Meaningful gamification and students’ motivation: a strategy for scaffolding
reading material. Online Learn. 22(2), 141–155 (2018)
29. Bye, R.T.: A flipped classroom approach for teaching a master’s course on artificial intelli-
gence. In: Escudeiro, P., Costagliola, G., Zvacek, S., Uhomoibhi, J., McLaren, B.M. (eds.)
CSEDU 2017. CCIS, vol. 865, pp. 246–276. Springer, Cham (2018). https://doi.org/10.1007/
978-3-319-94640-5_13
30. Song, Y., Jong, M.S.Y., Chang, M., Chen, W.: Guest editorial: “HOW” to design, implement
and evaluate the flipped classroom? – a synthesis. Educ. Technol. Soc. 20(1), 180–183 (2017)
31. Çakiroglu, Ü., Öztürk, M.: Flipped classroom with problem based activities: exploring self-
regulated learning in a programming language course. J. Educ. Technol. Soc. 20(1), 337
(2017)
32. Fernández, A.R., Merino, P.J.M., Kloos, C.D.: Scenarios for the application of learning ana-
lytics and the flipped classroom. In: 2018 IEEE Global Engineering Education Conference
(EDUCON), pp. 1619–1628. IEEE (2018)
33. Jovanovic, J., Gaševic, D., Dawson, S., Pardo, A., Mirriahi, N.: Learning analytics to unveil
learning strategies in a flipped classroom. Internet High. Educ. 33, 74–85 (2017). https://doi.
org/10.1016/j.iheduc.2017.02.001
34. Blau, I., Shamir-Inbal, T.: Re-designed flipped learning model in an academic course: the role
of co-creation and co-regulation. Comput. Educ. 115, 69–81 (2017)
35. Kaw, A., Clark, R., Delgado, E., Abate, N.: Analyzing the use of adaptive learning in a flipped
classroom for preclass learning. Comput. Appl. Eng. Educ. 27(3), 663–678 (2019)
154 M. Algayres and E. Triantafyllou
36. Yang, Y., Wu, H., Cao, J.: Smartlearn: predicting learning performance and discovering
smart learning strategies in flipped classroom. In: 2016 International Conference on Orange
Technologies (ICOT), pp. 92–95. IEEE (2016)
37. Hwang, G.J., Chen, P.Y.: Effects of a collective problem-solving promotion-based flipped
classroom on students’ learning performances and interactive patterns. Interact. Learn.
Environ. 1–16 (2019). https://doi.org/10.1080/10494820.2019.1568263
38. Ji, Y., Han, Y.: Monitoring indicators of the flipped classroom learning process based on
data mining-taking the course of “virtual reality technology” as an example. Int. J. Emerg.
Technol. Learn. 14(3), 166–177 (2019)
39. Valdez, M.T., Ferreira, C.M., Barbosa, F.M.: Implementation of methodological strategies,
attitudes and instruments as a PBL resource. In: 2018 17th International Conference on
Information Technology Based Higher Education and Training (ITHET), pp. 1–4. IEEE
(2018)
Distributed Student Team Work
in Challenge-Based Innovation
and Entrepreneurship (I&E) Course
1 Introduction
Challenge-based education is alternative to traditional education in training
engineering graduates to become independent learners, critical thinkers, prob-
lems solvers, life long learners as well as team payers [16]. This educational model
is relatively (dates from around 2011), it is built on problem-based learning, and
it represents the next step forward. It does not require only to have a problem
at the center of the learning process, it requires the problem proposer to be
c Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 155–163, 2020.
https://doi.org/10.1007/978-3-030-38778-5_18
156 G. Pisoni et al.
In our paper we present one such challenge-based course and investigate how
students physically located at distant Universities can collaborate and work in
team between each other in it. We present how we designed the course and
the results we obtained from questionnaires delivered after the course asking
them questions on which aspects of the experience they appreciated the most
and why, which aspects students appreciated less and why, if they liked the
experience and if they found it useful. The cross-university collaboration that
we describe in this paper took place between University of Trento, UNITN, Italy
and Universidad Politécnica de Madrid, UPM, Spain over the course Innovation
and Entrepreneurship Study delivered in both of the places at the same time.
This paper is structured as follows: Sect. 2 presents the educational context,
the course structure, and the methods we used to conduct this study, Sect. 3
presents the results from the evaluation with the students and the impact of the
course on their learning, in Sect. 4 we discuss the leanings from the deployment
and in Sect. 5 we outline our next steps.
2.3 Method
After completion of the course the students part in the mixed groups under-
took questionnaire evaluation. All the six students participants in the mixed
teams responded.
Below a short description on the challenges that the mixed teams worked on:
Challenge on Stand-alone Voice Assistants (provided by Comunica A+):
From the perspective of an advertising agency and marketing company such as
Comunica + A, and the technology of Voice Assistants, the company is interested
to understand: how will the customers’ relationship and sale models change?
What value-added services can an agency provide using this technology? How
will be the monetization model for the technology providers? Will a unique tech-
nology/provider monopolize the market? And where should the company start
from and where to aim?
Distributed Team Work in an I&E Course 159
Students Considered the Pilot a Useful Learning Experience. Some of them com-
mented that it was nice to work with different cultures over long distance, and
that it is something that will most likely be a part of the working life later on.
One students commented: “The team work was challenging having to coordinate
remotely but at the same time it was a fun experience doing all the brainstorming
sessions over Skype meetings and developing over our work”.
Another student commented that it was great to learn how to manage group
projects remotely, and that the only minor downside was not directly meeting
the company representative. Indeed this is an aspect which we need to improve
in our future design and think thoroughly how to provide the same “access” to
the challenge provider to all the students in the teams.
In this respect, and how to improve communication with the distributed
challenge provider, one of our students suggested: “I think there could be one
or two personal meet ups organized between team members and the industry so
that the team members away from the country of the industry location also get
a chance of exposure to what they are actually dealing with”.
Students Considered That the Pilot Increased Their Sense to the Community.
One of them commented: “Get to know (or know better) people that are part
of the community but have chosen a different university path”. Another student
shared that it was interesting to learn how it hear the experience on how it is to
study at the other university where this participant did not studied yet.
Students Felt That the Pilot Increased Their Ability to Work in Distributed
Teams. The general comments were that the fact that the team members were
not physically close to them made them more responsible in the process of plan-
ning the video calls and were more sensitive for the availability of the other
team members. One student commented: “Scattered chances of meeting meant
learning how to properly manage those kind of workload”, while another student
shared that “Collaborating over Skype is a different skill than discussing things
in person, and I feel that this skill has been improved”.
Students Were Engaged in Learning. One students commented that in the begin-
ning the engagement was low due to the distance with the company, however
as the assignment was getting more clear, the engagement level significantly
improved: “Not a lot at the beginning since the request from the company was
fuzzy, but greatly increased going on”.
Technology Support Was Average and Could be Further Improved. Students both,
acknowledged and criticized the support the technology was providing in their
team work. One student for instance commented: “Couldn’t have done it without
video conference, screen sharing and editing tools like Google docs (specially for
the revision system)”, while other said that “We had a terrible wifi connection
at home, so we could not use video and the phone call would break up a lot
of times. Moreover, we did not have time slots free during the week so we had
to Skype during the night, on which times the wifi was worse than during the
day”. The common problems reported were mainly of a technological nature and
Distributed Team Work in an I&E Course 161
4 Conclusions
Our aim with this paper was to develop innovative teaching format for students
to get deeply involved with company provided cases using a participative and col-
laborative approach connecting the two Universities involved. The multi-method
teaching approach based on group projects increased student’s ICT skills as well
as their ability to work in heterogeneous groups.
The experiment was successful in terms of operation and in terms of satis-
faction of the students and the companies. We also observed that the students
in the mixed teams, despite the difficulties of the physical separation, were more
motivated to work and that they had expanded their network of contacts within
the EIT community.
On the other side, the experiments forced the faculty of both universities to
coordinate and share practices, enriching the original version of their own I&E
Study course. In the experiment we also noticed that the I&E Study case, with
all the interaction with the company provider and the final exposition of results
162 G. Pisoni et al.
5 Next Steps
In future we want to repeat and extend the concept of the experiment in 2019. We
will include in 2019 students from University of Aalto, with the aim of including
other universities from the EIT Digital network in 2020. Based on our experi-
ence from this year, the organisers founding the experience too burdensome to
coordinate distributively, we will try a model in which there is only one univer-
sity organizing the course and the others are just users. Learning also from the
experience that the coordination and collaboration in teams for the students was
not easy in distributed settings as ours, and with the aim to even further study
collaborative learning dynamics, so we will incorporate in the teams’ dynamics
the use of tools to make their coordination and communication easier: Slack
for communication and Trello for planning and coordination of tasks. Finally,
developing of communication and entrepreneurship skills require not only learn-
ing from experts but also from peers, so we will include in the final pitching a
peer-evaluation phase made by all students using a voting tool such as Voxvote.
The course will be developed as follows: (1) UPM will play the role of Organizer.
The organizer will be in charge of supervising each case, mentoring the teams.
UNITN and AALTO will play the role of User, with their students remotely
supervised by the organizer. (2) Organizer and Users will synchronize lectures,
online modules and assignments (3) The students from the three universities will
be assigned to cases, in teams composed by at least one member from each uni-
versity. (4) In any team the students from the country of origin of the case will
be in charge of communication with the company provider. (5) The students will
plan and develop the case using professional tools, mentored by the organizer.
(6) There will be a final “entrepreneur pitching” evaluated by the “investors
committee” (teachers and company), and peers.
Acknowledgment. The authors would like to acknowledge and thank the companies
(Graffter, Comunica A+, Minsait - Indra) for their participation and all the students
who took part in this ‘cross-university’ pilots initiative. The pilots were supported
financially by EIT Digital under the Innovation and Entrepreneurship Improvements
Education projects.
References
1. Abrami, P.C., Bernard, R.M., Bures, E.M., Borokhovski, E., Tamim, R.M.: Inter-
action in distance education and online learning: using evidence and theory to
improve practice. J. Comput. High. Edu. 23(2–3), 82–103 (2011)
2. Boling, E.C., Hough, M., Krinsky, H., Saleem, H., Stevens, M.: Cutting the dis-
tance in distance education: perspectives on what promotes positive, online learn-
ing experiences. Internet High. Edu. 15(2), 118–126 (2012)
Distributed Team Work in an I&E Course 163
3. Charosky, G., Leveratto, L., Hassi, L., Papageorgiou, K., Ramos-Castro, J., Bragós,
R.: Challenge based education: an approach to innovation through multidisci-
plinary teams of students using design thinking. In: 2018 XIII Technologies Applied
to Electronics Teaching Conference (TAEE), pp. 1–8. IEEE (2018)
4. Crepon, R.: Closing the university-business gap: model of blended learning for the
collaborative design of learning resources
5. Dion, G., Dalle, J., Renouard, F., et al.: Change management: blended learning
adoption in a large network of European universities. In: International Conference
on e-Learning, pp. 77–83. Academic Conferences International Limited (2018)
6. Faltin, N., Böhne, A., Tuttas, J., Wagner, B.: Distributed team learning in an
internet-assisted laboratory. In: International Conference on Engineering Educa-
tion, pp. 18–22, Manchester (2002)
7. Pisoni, G., Hegyi, B., Marchese, M., Renouard, F.: Portfolio of innovative online
courses in a Pan-European network of universities. In: INTED (2019)
8. Kirch, J., Eisenbart, B.: Teaching entrepreneurship, digitalization, leadership, and
gender at the same time: how a new learning approach integrates all four perspec-
tives. In: ICERI (2018)
9. Malmqvist, J., Rådberg, K.K., Lundqvist, U.: Comparative analysis of challenge-
based learning experiences. In: Proceedings of the 11th International CDIO Con-
ference, Chengdu University of Information Technology, Chengdu, Sichuan, PR
China (2015)
10. Oztok, M., Zingaro, D., Brett, C., Hewitt, J.: Exploring asynchronous and syn-
chronous tool use in online courses. Comput. Edu. 60(1), 87–94 (2013)
11. Pisoni, G., et al.: Towards blended learning implementation of innovation and
entrepreneurship (I&E) education within eit digital: the models and lessons learnt.
In: EDULEARN18 Proceedings of 10th International Conference on Education and
New Learning Technologies, IATED, pp. 10496–10502, 2–4 July 2018. https://doi.
org/10.21125/edulearn.2018.2553
12. Pisoni, G., Marchese, M., Renouard, F.: Benefits and challenges of distributed
student activities in online education settings: cross-university collaborations on
a Pan-European level. In: 2019 IEEE Global Engineering Education Conference
(EDUCON), pp. 1017–1021 (2019)
13. Pisoni, G.: Strategies for pan-european implementation of blended learning for
innovation and entrepreneurship (I&E) education. Edu. Sci. 9(2), 124 (2019)
14. Ramirez-Mendoza, R.A., et al.: Towards a disruptive active learning engineering
education. In: 2018 IEEE Global Engineering Education Conference (EDUCON),
pp. 1251–1258. IEEE (2018)
15. Serçe, F.C., Swigger, K., Alpaslan, F.N., Brazile, R., Dafoulas, G., Lopez, V.:
Online collaboration: collaborative behavior patterns and factors affecting globally
distributed team performance. Comput. Hum. Behav. 27(1), 490–503 (2011)
16. Willis, S., Byrd, G., Johnson, B.D.: Challenge-based learning. Computer 50(7),
13–16 (2017)
17. Yoo, Y., Boland Jr., R.J., Lyytinen, K., Majchrzak, A.: Organizing for innovation
in the digitized world. Organ. Sci. 23(5), 1398–1408 (2012)
Semantic Recommendations and Topic
Modeling Based on the Chronology of Romanian
Literary Life
1 Introduction
In the context of understanding the importance of Romanian authors and writings across
time, one of the most important on-going projects of the “G. Călinescu” Institute of
Literary History and Theory is the Chronology of Romanian Literary Life (commonly
referred to as CVLR). CVLR is aimed to register chronologically all the important
literature-related events which happened nationwide after World War II.
Two important periods are mapped from the historical perspective, namely commu-
nism (years between 1949 and 1989) and post-communism (years between 1990 and
2000; year 2000 being the last one which is covered in the current version of CVLR).
The total number of years to be documented by the chronology is 56, whereas the cur-
rent analysis considers only 24 years, the remaining ones being work in progress or not
having a proper format for parsing and indexing.
Our aim is to capture the evolution of topics across time and to recommend literary
events from the CVLR considering searches on keywords mapped to specific topics.
Our work relies on Natural Language Processing (NLP) and Information Retrieval (IR)
techniques tailored for our aim [1]. Moreover, one of the main areas of interest in the
domain of AI in Education is related to providing recommendations of learning objects
and adaptive learning support [2], which is currently a challenging task. Even though the
range in which recommender systems are used is widen (from which products to buy, to
which music to listen, or which information to learn from) [3], the end goal is the same:
to help users or learners to take decisions easier.
The proposed solution aims to provide relevant recommendations of historical liter-
ary events to the learner, which are based on previously discovered topics of interest. To
our knowledge, this is the first work which analyzes, summarizes, and presents the topics
of interest, their evolution in the history of the Romanian literary life. and is able to make
recommendations of historical events which can be used by the learning communities
interested in Romanian literary.
Chapter two describes similar work conducted in the Information Retrieval field, the
usage of semantic models, and discusses how similar recommender systems were imple-
mented and used in education. The method chapter presents the used corpus, how the
data was pre-processed, details on topic modeling and determining the topic coherence
score, as well as recommender system was developed. The paper continues with results
obtained and it is followed by conclusions and possible future work directions.
Topic modeling considers NLP techniques used to detect the latent topics of a docu-
ment. Latent Dirichlet Allocation (LDA) [4] is a probabilistic graphical model used to
automatically find the latent topics. In this model, both the distribution of topics and the
distribution of words over a topic are assumed to have a sparse Dirichlet prior distribu-
tion, meaning that each document is assumed to be composed only of a small number
of topics, and each topic is assumed to be formed mainly on a limited number of words.
are recommended. There are several techniques used to create a user profile and to gather
information, such as questionnaires, users’ item ratings through time (which are called
explicit techniques), or learning based on the users’ activities through time (which are
the implicit techniques).
Thus, a recommender system is part of one of the three categories listed below based
on the methods used for gathering information [2]:
Even though the recommender systems were initially used and are popular in other
fields, such as in recommending movies, news or e-commerce, there are more and more
scenarios in which such systems are used in education, more specifically in the e-learning
environments.
There are still some issues, which are specific for e-learning recommender systems,
and those are [6]: (a) items which are marked of interest for learners might not be
pedagogically appropriate for them; (b) customization should not only be made about
the choice of learning items, but also about their delivery; and (c) learners are not expected
to read too many documents/information.
Several approaches on e-learning recommender systems were conducted, out of
which some were aimed to recommend online learning activities or shortcuts in a
web course based on learners history using web mining [7]. Dagadita et al. [3] rec-
ommended articles based on content-based filtering and collaborative filtering, whereas
Klasnja-Milicevica [8] personalized programming curricula for learning using hybrid
recommendation strategy.
3 Method
Our proposed solution is a content-based recommender system, storing a user profile
based on the items user has searched and marked them of interest in the past. The
data queried is the Chronology of the Romanian Literary Life and each event in the
chronology is associated to a set of topics. An item in the recommender system is a
particular event and, based on user preferences of items, system can map topics which
user likes and is able to make recommendations of similar events.
Figure 1. depicts the processing pipeline, where CVLR corpus was initially indexed
in Elasticsearch and data was afterwards pre-processed. After the pre-processing step,
LDA is applied in order to identify trending topics. The last step in the processing pipeline
consists of presenting possible use-cases of our method in real scenarios, where learners
can search for literary events using keywords and the system is able to recommend
further events based on their profile.
Semantic Recommendations and Topic Modeling 167
•Indexing in Elasticsearch
1 •Data Pre-processing
•Topic modeling
2
3.1 Corpus
The corpus consists of events indexed chronologically by day. Each event generally
describes a publication of an author, some of them containing a short description of the
impact of the publication in the literature field. As previously mentioned, not all years
are covered in our analysis, because some of them were not yet written or could not be
properly parsed. Thus, 24 years are covered: 1949–1959, 1964–1967 and 1990–2000,
with the exception of years 1957 and 1999. Each event is considered a separate document
in the following analyses; the current work relies on 1962 indexed entries.
Stop words were filtered for each document, together with words which contained non-
alphabetic characters and words with a high frequency (i.e., words appearing in more
than 10% of the documents), or words with an absolute frequency lower than 4. For each
remaining word, only the lowercase version was considered. There was also a process
of grouping neighboring words within the same named entity category as these words
are referring to the same entity. The NER (named entity recognition) was trained on
RONEC corpus1 with a standard spacy model for NER2 .
After the subsequent pre-processing operations are performed, LDA is applied. More-
over, specific methods which approximate LDA parameters must be applied because the
exact inference of the parameters is considered intractable. The genism LDA model3 was
used which implements an online variational Bayes algorithm [9]. LDA cannot infer the
number of topics; therefore, multiple LDA models were trained with different number of
topics and the coherence score was computed for each of them, followed by the selection
1 https://github.com/dumitrescustefan/ronec.
2 https://spacy.io/usage/training.
3 https://radimrehurek.com/gensim/models/ldamodel.html.
168 L.-M. Neagu et al.
of the number of topics that maximizes coherence (in our case 20 topics). The coherence
score is computed using the genism4 which implements the algorithm introduced by
Röder, Both and Hinneburg [10]. The cohesion values depending on incremental topic
numbers is depicted in Fig. 2.
0
-0.5 0 20 40 60 80 100 120 140
Coherence
-1
-1.5
-2
-2.5
-3
Number of topics
importance. After reading an event in particular, based on its ID, users are able to mark
it of interest and store it in their preferences.
The storing mechanism relies on an XML file whose root node has children contain-
ing user IDs. For each child, there is a list of other children which are the event ids. On
when showing interest for a specific event, a new entry is appended in the XML file to
that specific user ID parent. Within the reversed process, that entry is removed from the
file, if it exists.
A simple user profile/learner model is built by using the preferred events by each
specific user. In a typical e-learning task recommender system [7], the recommendations
are based on the tasks already done by the learner and the measure of their success,
but also on tasks made by “similar” learners. Common techniques used for providing
recommendations in the technology enhanced learning field consider [11]:
As the current system stores the user preferences, it is possible to provide recommen-
dations based on their topics of interest. In order to achieve this, the system checks if
the current user has any preferred events and, if not, it recommends the events with
the highest value for each topic. If users have previously stored into their profile a list
of events, the system maps their events of interest with corresponding topics and new
events are recommend based on those topic scores.
4 Results
Elasticsearch was used to store the CVLR corpus, whereas LDA modeling, search engine
and recommendations functionalities were developed in Python, with no GUI yet linked
to the backend functionalities. The obtained results are interesting. The first subsection
introduces the trending topics of the Romanian literary life after World War II and their
evolution, year by year. In the second part, the search engine is described, including
topic modeling results and the recommender system, following a simple use-case with
corresponding step-by-step results.
After performing the topic modeling, the list of 20 top topics in the corpus, with their
associated words and weight was obtained. Topics, numbered from 1 to 20, together with
their top 7 words and corresponding theme are presented in Table 1. In some cases, the
same word appears in the same topic in different forms (e.g., “socialist” and “socialism”
in topic 2) due to automated lemmatization errors. Another problem is the appearance
of different words describing the same entity (e.g., “Paul Goma” and “Goma” in topic
170 L.-M. Neagu et al.
with ID 1). Although named entities were merged, the authors are not mentioned in the
same manner all the time; therefore, all versions may still be assigned high probabilities
by the LDA.
The reason behind having different words naming the same concept is generated by
the manner in which CVLR was written, namely by the usage of artistic literary language
which is different than standard Romanian language. Within artistic literary language,
authors use personalized stylistic methods for creating their texts.
After LDA was trained on the entire set of events, an importance to each topic was
assigned for every year, and then sorted by importance. This assignment is performed
as the sum of topic probabilities for each event in that year. Thus, the evolution of the
importance of these topics is shown in Fig. 3. The missing years are substituted with a
linear function based on the known neighboring years. Only 7 emblematic topics were
selected for this visualization; the remaining topics showed a similar trend to one of
those selected.
Interesting trends are captured by the topic evolution analysis. In the beginning of
the 1950s, the most prominent topic has ID 2, which is mainly about socialism (frequent
words are “muncitor”, “socialist”, “clasă”, “erou”, etc.). This is due to the influence
of the Russian Soviet Union which aimed to create “the new human”, aligned with
their socialist vision and started with cultural press propaganda, driven mainly by young
writers or writers with a left-party vision. However, its importance drops significantly at
the beginning of the 1960s, which indicates a moment of change in the relationships with
the Russian Soviet Union (known as the moment of “cooling of brotherly relations”,
started by Gheorghe Gheorghiu-Dej and continued by president Nicolae Ceaus, escu).
Although it is still the communist era, the literature field distances itself from communism
at the beginning of 1960s. In this period, events shift towards topics 16, 20 and 11 which
are focused on literature; this announces a new generation, with European and Global
literary values, and also the moment when part of the Romanian great writers returned
from the Exile. In the 1990s, the main topics are 6, 12 and 1. Topic 1 and 6 contain a lot
of words related to dissidence in the communist period. For instance, topic 1 refers to the
author Paul Goma, an author well-known for his critique of the communist, and which
generated many debates on recovering our literary past values. Therefore, communism
dissidence is highly present in the post communism events.
600
Coherence score
400
200
other events within the same topics, based on the previously entries. A sample of this
flow is described below. For example, the user searches for the following keywords (1):
The search query is initially filtered using spacy model trained for Roma-
nian Language from the ReaderBench Python library, and the filtered query is the
following (2):
The final list of keywords (only content words) is afterwards passed on to each
topic, and a corresponding weight is computed. Afterwards, the weights are ordered in
descending order, and only the relevant topics are kept. In the list of top topics, the weight
percentage of each topic is calculated, and a maximum number of events is returned. For
the provided example, the corresponding values for top topics are displayed in Table 2.
Topic Id Weight Weight percentage Events no. to display Top words in topic
18 .0476 64.99% 6 premiu, leu, juriu, acorda,
premiat
4 .0126 17.2% 2 pământ, patrie, inimă,
munci, sat
19 .0067 9.19% 1 sat, dezvoltare, socialist,
copil, nuvelă
2 .0063 8.61% 1 muncitor, socialist, clasă,
realism, socialism
As it can be observed, the most prominent topic within the provided list of keywords
is topic 18, which has as top words some of the words provided directly in the search
query: “premiu” (eng. “prize”) and “juriu” (eng., “jury”). Also, results are similar for
the other returned topics, which have top words similar to the provided keywords. The
final results are displayed as ordered events (based on their weight, in descendent order)
within each topic. One of the most relevant returned events is the following (3):
Users are also able to store events as favorites in their profile. After building it,
users can query for recommendations based on the events they are interested in. In this
Semantic Recommendations and Topic Modeling 173
case, the system checks for potentially relevant events and a weight for each topic is
calculated. Afterwards, a strategy similar to the one employed by the search engine is
used to provide personalized recommendations: topics are ordered descendent by weight,
a weight percentage is calculated for each topic, and a number of events, in this case
excluding the ones already stored by the user, is retrieved.
Shifting the Romanian literature to the digital era is an ambitious effort which includes
multiple approaches and methods. This paper introduces NLP and IR techniques in order
to determine the emerging topics from the Chronology of Romanian Literary across time.
Moreover, we propose a solution to improve the search for events in CVLR, together
with semantic recommendations. The current work shares similar goals with previously
performed analyses on the General Romanian Dictionary of Literature [12].
Future work will be conducted to include the missing years from CVLR. In addition,
the recommender system will be enhanced by taking into account further details about
users, their search history, and a more comprehensive profile in order to provide more tai-
lored recommendations. Moreover, designing an intuitive user interface and publishing
the search engine online is one of our main priorities.
Acknowledgements. This work was supported by a grant of the Romanian National Authority
for Scientific Research and Innovation, CNCS – UEFISCDI, project number PN-III 54PCCDI /
2018, INTELLIT – “Prezervarea s, i valorificarea patrimoniului literar românesc folosind solut, ii
digitale inteligente pentru extragerea s, i sistematizarea de cunos, tint, e”.
References
1. Manning, C.D., Raghavan, P., Schütze, H.: Introduction to Information Retrieval, vol. 1.
Cambridge University Press, Cambridge (2008)
2. Persen, W.C.A.R.: A Recommender System for Collaborative Knowledge. books.google.com
1 (2009)
3. Dagadita, M., Bancu, C., Dascalu, M., Dobre, C., Trausan-Matu, S., Florea, A.M.: ARSYS
- article recommender system. In: 14th International Symposium on Symbolic and Numeric
Algorithms for Scientific Computing (SYNASC 2012). IEEE, Timisoara (2012)
4. Blei, D.M., Ng, A.Y., Jordan, M.I.: Latent dirichlet allocation. J. Mach. Learn. Res. 3(4–5),
993–1022 (2003)
5. Ricci, F., Rokach, L., Shapira, B., Kantor, P.B.: Recommender Systems Handbook, p. 845.
Springer, Boston (2011). https://doi.org/10.1007/978-0-387-85820-3
6. Tang, T.Y., McCalla, G.: Smart Recommendation for an Evolving E-Learning System:
Architecture and Experiment. Int. J. E-learn. 4, 105–129 (2005)
7. Zaiane, O.R.: Building a recommender agent for e-learning systems. In: International
Conference on Computers in Education, Proceedings, p. 5 (2002)
8. Klasnja-Milicevica, A., Boban Vesina, M.I., Budimac, Z.: E-Learning personalization based
on hybrid recommendation strategy and learning style identification. Comput. Educ. 56, 885–
899 (2011)
174 L.-M. Neagu et al.
9. Hoffman, M., Bach, F.R., Blei, D.M.: Online learning for Latent Dirichlet Allocation. In:
Advances in Neural Information Processing Systems, p. 10 (2010)
10. Röder, M., Both, A., Hinneburg, A.: Exploring the space of topic coherence measures. In:
Proceedings of the Eighth ACM International Conference on Web Search and Data Mining,
pp. 399–408. ACM (2015)
11. Manouselis, N., Drachsler, H., Vuorikari, R., Hummel, H., Koper, R.: Recommender Systems
in Technology Enhanced Learning. Springer, New York (2014). https://doi.org/10.1007/978-
1-4939-0530-0
12. Neagu, L.-M., Cotet, T.-M., Dascalu, M., Trausan-Matu, S., Badescu, L., Simion, E.: Semantic
author recommendations based on their biography from the General Romanian Dictionary of
Literature. In: 7th International Workshop on Semantic and Collaborative Technologies for the
Web, in Conjunction with the 15th International Conference on eLearning and Software for
Education (eLSE 2019), pp. 165–172. “CAROL I” National Defence University Publishing
House, Bucharest (2019)
A Web-Based Platform for Building PBL
Competences Among Students
1 Introduction
At Aalborg University, the degree programmes are based around problem-based
learning, and in most cases this takes the form of a project-based approach. Stu-
dents use a variety of Internet-based infrastructures for sharing their experiences
with PBL projects, and research results indicate that students generally infor-
mal infrastructures originally intended for other purposes, such as Facebook and
Pinterest, for many of their PBL activities [8]. This informal digitalization of
PBL may be beneficial for the project process but it also leads to risks. In par-
ticular, informal traditions and ways of seeing the learning process may emerge
that lead to a form of ritualization that is detrimental to actual learning [1].
Another challenge is that the sharing of experience only happens within well-
defined communities, usually among students within the same degree programme
or even within the same year. This is despite the fact that many experiences are
similar across degree programmes and years.
c Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 175–182, 2020.
https://doi.org/10.1007/978-3-030-38778-5_20
176 H. Hüttel et al.
There are many different ways of viewing the same kinds of challenges in the
project process, but most students never become aware of the fact that many
students from different degree programmes face the very same problems that
they face and may be able to contribute to their understanding of the situation
through a different perspective.
In this paper we describe the PBL Exchange/stud platform and our experience
with it. The goal of this web-based platform has been to support student pro-
cesses of reflections on and sharing of experiences with PBL. PBL Exchange/stud
is a structured space in which communication can be carried out by students
and in which experience can be shared across years and degree programmes and
achieve a form of permanence. Moreover, the goal has been to involve lecturers
and project supervisors through guided activities. PBL Exchange/stud is inspired
by previous work on a platform for PBL supervisors [2]. A major difference is
that activity within PBL Exchange/stud is to be initiated through interventions
from teaching staff.
In the rest of this paper we describe the central features of the PBL
Exchange/stud platform, our use of it through guided interventions and our reflec-
tions on the outcome of these.
Fig. 1. Screenshot of the PBL Exchange/stud user interface. The interface lists the most
recent questions asked and shows the scores (in pale green boxes). (Color figure online)
– ask questions
– answer questions
– comment answers and questions that are already in the system
– upvote and downvote questions, answers in the system
Most importantly, PBL Exchange/stud must be able to interface with other sys-
tems related to teaching activities at Aalborg University – in particular with the
main platform Moodle [4]. All services share a common entry point for authen-
tication in the form of a single sign-on feature.
As part of the joint effort with PBL Exchange meant for PBL supervisors,
we decided to re-implement a version of PBL Exchange/stud using the Python-
based Django framework. This framework was chosen because it can help ensure
plugability and ease of reusability of Python code. This allows us to incorporate a
collection of new features, including multilingual support and mail notifications.
Using the Django framework allowed us to build the structure around mod-
ules, where we for modules such as the authentication module, could use exist-
ing modules that had already been tested. The system architecture of PBL
Exchange/stud showing the four main modules is shown in Fig. 2. The four mod-
ules are written to be independent. Some functionality of each module may
require another module to be active, but this is not required. An example of this
is the questions module, which only activates the scoring system if the users mod-
ule is present. The PBLExchange module forms as the base of the site, defining
the front page and the base layout that the other modules extend.
4 The Interventions
The challenge of developing, maintaining and extending competences within
PBL exist both at the undergraduate and postgraduate level; postgraduate stu-
dents in Denmark may come from other institutions that have no tradition for
using problem-based learning. For this reason we decided to study how the PBL
Exchange/stud platform could be useful in the setting of teaching activities whose
stated goal has been to get students to develop PBL competences.
A Web-Based Platform for Building PBL Competences Among Students 179
Moreover, at least in the case of first-year students, it was not obvious to them
why they should give priority to answers from their fellow students when they
had experienced members of teaching staff to draw on. Therefore, scaffolding
should seek out situations where students would naturally want to learn from
the experience of other students. Still, students could be asked to systematically
document and reflect on their experiences with PBL Exchange/stud, and if one
made sure that students got credit for such an activity, this could be an extrinsic
motivation.
References
1. Hüttel, H., Gnaur, D.: If PBL is the answer, then what is the problem? J. Probl.
Based Learn. High. Educ. 5(2), (2017). https://doi.org/10.5278/ojs.jpblhe.v5i2.
1491
2. Hüttel, H., Gnaur, D., Hairing Klostergaard, A., Blegmand, G.: A new platform for
question-based sharing of supervision competencies in problem-based learning. In:
Rodrigues, A., Fonseca, B., Preguiça, N. (eds.) CRIWG 2018. LNCS, vol. 11001, pp.
177–184. Springer, Cham (2018). https://doi.org/10.1007/978-3-319-99504-5 14.
http://dblp.uni-trier.de/db/conf/criwg/criwg2018.html#HuttelGKB18
3. McComas, W.F.: “21st-Century Skills”, p. 1. SensePublishers, Rotterdam (2014).
https://doi.org/10.1007/978-94-6209-497-0 1
4. Moodle. http://www.moodle.org
182 H. Hüttel et al.
5. NEA: Preparing 21st century students for a global society: an educator’s guide to
the “four cs” (2012). http://www.nea.org/tools/52217.htm
6. Question2Answer. https://www.question2answer.org
7. Ryberg, T., Davidsen, J.: Establishing a sense of community, interaction, and
knowledge exchange among students. In: Kergel, D., Heidkamp, B., Telléus,
P., Nowakowski, S., Rachwal, T. (eds.) The Digital Turn in Higher Education.
Springer, Wiesbaden (2018). https://doi.org/10.1007/978-3-658-19925-8 11
8. Ryberg, T., Sørensen, M.: Sociale medier i undervisning og uddannelse: poten-
tialer og problematikker. Læring og Medier 12(21), 1–11 (2019). https://doi.org/
10.7146/lom.v12i21.112894. (in Danish)
9. Ryberg, T., Wentzer, H.: Erfaringer med e-porteføljer og personlige læringsmiljøer.
Dansk Universitetspaedagogisk Tidsskrift 11, 14–19 (2011)
10. Sørensen, M.T.: The students’ choice of technology a pragmatic and outcome-
focused approach. In: Kergel, D., Heidkamp, B., Telléus, P., Rachwal, T.,
Nowakowski, S. (eds.) The Digital Turn in Higher Education. Springer, Wiesbaden
(2018). https://doi.org/10.1007/978-3-658-19925-8 12
11. Stackexchange. http://www.stackexchange.com
12. Uskoković, V.: Flipping the flipped: the co-creational classroom. Res. Pract. Technol.
Enhanc. Learn. 13(1), 11 (2018). https://doi.org/10.1186/s41039-018-0077-9
13. Wenger, E.: Communities of Practice: Learning, Meaning, and Identity (Learn-
ing in Doing: Social, Cognitive and Computational Perspectives), pbk. edn.
Cambridge University Press (1999). http://www.amazon.de/Communities-
Practice-Cognitive-Computational-Perspectives/dp/0521663636/ref=wl it dp v?
ie=UTF8&coliid=IDPFVX4U6T022&colid=1SYIKSJ6TTEE3
Reconstructing Scanned Documents
for Full-Text Indexing to Empower Digital
Library Services
slvtomescu@gmail.com
Abstract. The digital era raises new challenges for traditional library services
in which information has to be delivered and supported by technology-enhanced
systems. The increasing need for rapid access to information requires librarians
to re-evaluate the way they develop, manage and deliver resources, as well as ser-
vices. However, most information extraction systems are not designed to work with
PDF files generated after Optical Character Recognition, and several problems are
encountered while trying to properly restructure the recognized text, for example:
disruption of paragraphs, improper page breaks, or loss of content structure. This
paper introduces a pre-processing pipeline designed to support university libraries
to adequately index old document collections. The extracted text is indexed into
Elasticsearch which facilitates the search for relevant documents, based on key-
words. The information extraction system is designed to assist librarians in the
digitization process by enabling a systematic review of documents, which leads
to more accurate representations of the indexed files.
1 Introduction
The digital services provided by libraries have emerged from the need to easily retrieve
relevant information from the wide range of existing physical and electronic documents.
Our project is aimed to facilitate the digitization process of the Central University Library
of Bucharest, which currently hosts over 2 million physical volumes [1], for information
retrieval purposes and integration in an e-learning environment. In the current global
context driven by advanced technologies [2], it is recognized that data preprocessing is
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 183–190, 2020.
https://doi.org/10.1007/978-3-030-38778-5_21
184 M. Nitu et al.
a critical part of any text mining, Natural Language Processing (NLP) and Information
Retrieval (IR) system.
This paper proposes an implementation of extraction mechanisms from unstructured
file formats, data indexing, and search for relevant documents based on keywords, stages
for which text processing techniques play a significant role. A processing pipeline con-
sists of several stages, each centered on a specific type of operation. Most information
extraction systems were not designed to work on Portable Document Format (PDF)
files in which the content is often mixed with publication metadata or semi-structured
text, thus introducing additional NLP challenges. Moreover, the absence of effective
means to extract text from these PDF files represents a struggle for researchers relying
on published literature as a primary source of information. We focus on the fundamental
text preprocessing stage, represented by an information extraction system that has the
potential to assist humans in the text extraction task, enabling a systematic review before
the indexing stage.
The following state-of-the-art section highlights existing approaches in terms of mod-
ern text processing techniques. The preprocessing stages and our pipeline are presented
in the Method section, starting with the used corpora, text parsing and preprocessing
algorithms, followed by document indexing. The obtained results are reported in the
following section, while conclusions and the roadmap of future work are outlined next.
2 Related Work
The growing need to efficiently find and extract information from documents led to the
development of various text mining tools. However, many of these tools were mainly
designed to extract information from text-based documents, like JSON and other formats
[3], while today a considerable part of the literature is published and distributed in PDF
format. The well-known Acrobat DC is one of the most widely used software for pdf
text extraction and processing. Despite its capabilities, the features are not free and there
is few freely available information on the employed algorithms.
In the following, we briefly present a selection of the most popular PDF extrac-
tion tools for comparison purposes, along with their features. Depending on the pre-
sented tool, the output format of the processed file can be TXT (pdftotext, LA-
PDFText, PdfMiner, PdfBox), XML (pdftohtml, pdftoxml, pdf2xml, PdfMiner, PDFEx-
tract, pdfXkt, pdf-extract, pdfx), HTML (pdftohtml, pdf2xml, PdfMiner, pdfXkt) or
JSON (Icecite [4]). Table 1 presents an overview of the above tools’ features. If a feature
is fully provided by the tool, it is represented with “X”; otherwise it is noted by “-”.
Most of the existing text processing tools were designed to extract text from formatted
PDF layouts. Neji framework [5] was built for biomedical concept recognition in journal
publications, while Maciocci [6] conducted a study on data extraction from scientific
manuscripts in formatted PDF files using computer vision techniques and developed
the ScienceBeam tool. Yet the challenge remains; the processing of non-formatted PDF
files, which is currently a popular research subject in the field.
Related approaches for text extracting features in unstructured PDF files were pre-
sented by Hassan & Baumgartner [7] who proposed three methods based on layout seg-
mentation for data extraction: conversion to a structured format, ontology-based wrap-
ping, and spatial reasoning. Other researchers approached the non-formatted files issue
Reconstructing Scanned Documents for Full-Text Indexing 185
(continued)
186 M. Nitu et al.
Table 1. (continued)
into stages: started with a layout analysis and continued with segmentation, character
and structure recognitions [8].
3 Method
3.1 Corpus
The current corpus consisted of 55 books in PDF format, written in Romanian language,
provided by Central University Library of Bucharest. The input data is represented by
a set of scanned books on which Optical Character Recognition (OCR) was performed.
The resulting PDFs do not follow a specific format, so we are working with a set of non-
formatted PDFs. There are many challenges when working with non-formatted PDFs.
As a result of the OCR phase, we have encountered different font sizes and font types
on the same page, in one text line or in one paragraph, and different styles for headers
and footers in the same book, which created difficulties in adjusting the preprocessing
pipeline algorithm.
Given an unformatted PDF book, the first step of the preprocessing pipeline algorithm
is the parsing phase, where the logical text blocks are being identified. The text extraction
can be a challenging task in unformatted files, as the text is rendered in different formats.
Determining the correct reading order of words is an essential step for the text extraction.
Words order within a line is easily correlated with the position of the words on the page,
while the order between lines can be difficult to identify on two-column layouts. This
challenge is addressed by using PdfBox which successfully handles reading order, the
translation of ligatures, and the characters with diacritics. The document is parsed on
lines, identifying relevant sections and metadata within the book: chapter titles, headings,
paragraphs of the body text content, images, tables, and the table of contents. Paragraph
boundaries are rebuilt, whereas hyphenated words are merged. The book title, the author,
and the publishing year are extracted from the first page (if present) and are passed to
the JSON object before the processing phase. In the processing workflow, the refined
text is sent to the user interface by converting the JSON object into editable HTML and
displaying it using TinyMCE customized text area, allowing the user to edit the text
before saving the file. The saved form of the processed text is converted to JSON and
sent to Elasticsearch for indexing.
We experimented several approaches to properly extract metadata. The first model
uses the table of content to identify the chapter titles and their associated pages, split-
ting the content into titles and paragraphs. In contrast, the second model identifies the
dominant font type and size on a page which is associated to the body content. Each
time a different font is encountered, the corresponding text is considered chapter title or
heading. A more detailed overview on the algorithms is presented in the next paragraphs.
The algorithm for table of content (TOC) parsing consists of three main steps: (a) the
identification of TOC’s first page, (b) pattern validation, and (c) reconstruction of TOC
boundaries. In order to identify the first page of the table of content, we start from the
premise that its title should match one of the keywords from our dictionary. However,
due to the condition of the original documents and as a result of Optical Character
188 M. Nitu et al.
Recognition phase, the words may contain white spaces or symbols or may be split on
multiple lines; this imposed additional challenges and an additional validation phase
using regular expression was introduced. As a general observation, most TOC lines in
our dataset are ending with digits representing the start page of the chapter. Therefore,
a condition for the pattern validation was created. Moreover, the TOC is found in the
first or last 10 pages of a book, which helped us create a heuristic for the identification
of TOC boundaries. Furthermore, TOC entries are parsed using a regular expression to
correctly extract the chapter title and the associated page range.
The second approach is used to identify chapters and paragraphs based on most
common font. The method relies on PDFTextStriper from pdfBox to strip the text,
ignoring its formatting. The text is analyzed line by line, based on font information
(font name and font size), as well as text position on the page. The data for each line is
stored in a list and the text is associated with body content or title content by comparing
the predominant font on the page, with each line. As a convention, the text having
a font smaller or equal in size than the most common font is considered part of a
paragraph. Because the dataset contains very old documents having unformatted layouts,
we encountered different fonts and different sizes within the same paragraph, within the
same line or even within the same word. Since our goal is to group the text content into
titles and paragraphs, the small font differences within a line are ignored, and only the
predominant font is considered.
After iterative experiments, the two models were combined into one robust tool
which improved individual performances by easily adapting to the structure of most
PDF books. Moreover, a mechanism for image extraction was implemented; the images
are saved into a local folder, and a tag is inserted in the place where the image was
extracted.
4 Results
Starting from the corpus of 55 unformatted PDFs, we disregarded 18 books with severe
problems while assessing accuracy (i.e., percentage of correctly extracted document
metadata, including table of contents); thus, we were left with a collection of 37 items.
The evaluation of the proposed pipeline on the processed books is detailed in Table 2:
18 books had a processing accuracy of 100%, 22 over 90%, 25 over 80%, and only 9
under 50%. The separation of document headings in the unformatted PDFs on a corpus
of 37 processed books had an overall precision of 78%, a recall of 75%, while the overall
F-score was 73%. The overall high accuracy of the PDF processing pipeline argues for
its applicability within the digitization process.
An in-depth overview on the causes of the problematic documents is presented in the
following section. Within the 18 disregarded books we identified the following issues:
Table 2. Count of documents having a statistical measure (Accuracy, Precision, Recall, F-score)
higher than the imposed threshold.
References
1. Biblioteca Centrala Universitara Carol I. http://www.bcub.ro/home/biblioteca-in-cifre/
biblioteca-in-cifre-la-31-decembrie-2018. Accessed 16 Aug 2019
2. Cervone, H.F.: Emerging technology, innovation, and the digital library. OCLC Syst. Serv. Int.
Digit. Libr. Perspect. 26(4), 239–242 (2010)
3. Schouten, K., Frasincar, F., Dekker, R., Riezebos, M.: Heracles: a framework for developing
and evaluating text mining algorithms. Expert Syst. Appl. 127, 68–84 (2019)
4. Korzen, C.: Icecite (2017). https://github.com/ckorzen/icecite. Accessed 16 Aug 2019
5. Santos, A., Matos, S., Campos, D., Oliveira, J.L.: A curation pipeline and web-services for
PDF documents. In: CEUR Workshop Proceedings, vol. 1650. http://ceur-ws.org/Vol-1650/
smbm16San-tos.pdf. ISSN 1613-0073
6. Maciocci, G.: ScienceBeam - using computer vision to extract PDF data. https://elifesciences.
org/labs/5b56aff6/sciencebeam-using-computer-vision-to-extract-pdf-data. Accessed 16 Aug
2019
7. Hassan, T.: Baumgartner, R.: Intelligent text extraction from PDF documents, pp. 2–6 (2005).
https://doi.org/10.1109/cimca.2005.1631436
8. Sasirekha, D., Chandra, E.: Text extraction from PDF document. In: IJCA Proceedings on
Amrita International Conference of Women in Computing, AICWIC, no. 3, pp. 17–19 (2013)
9. Ramakrishnan, C., Patnia, A., Hovy, E., Burns, G.: Layout-aware text extraction from full-text
PDF of scientific articles. Source Code Biol. Med. 7(1), 7 (2012)
Curating Educational Resources for Homework
Management: A Support Prototype
Abstract. Learning content curation plays an important role given the increasing
amount of educational resources available on the Web. The process implies search-
ing, collecting, annotating, filtering, organizing and sharing relevant resources for
a specific learning context. Our aim is to provide a support platform which allows
both teachers and students to become content curators, leveraging various levels
of expertise. More specifically, we propose a system dedicated to homework man-
agement, called EdReHo, which allows the collection and sharing of educational
resources needed to understand and solve assignments. When teachers create an
assignment in EdReHo, they can recommend also a set of resources relevant for
that topic, which are aimed to supplement the mandatory course material. The
students can also add useful resources and share them with peers, becoming more
actively involved in the process and benefitting from the “learning by searching”
approach. The paper describes the EdReHo system prototype in terms of concept,
features and implementation and illustrates its main functionalities.
1 Introduction
2 Related Work
In what follows, we discuss some relevant systems proposed in the literature, focusing
especially on retrieving, collecting, searching, organizing, tagging, rating and sharing
learning resources.
Paper [2] presents a platform for language learning in which educational resources
consist in links retrieved from social bookmarking sites; the search is done automatically
by the system, based on keywords attached by the teacher to each learning activity. When
accessing such a learning activity, the student is presented with a list of 10–15 links,
which they have the option to like or dislike; links are subsequently filtered based on the
number of votes received from the learners.
Another system that provides the option of searching resources through various
Web 2.0 services (such as YouTube, SlideShare, Blogger, Delicious etc.) is described in
[7]. LearnWeb2.0, as it is called, allows the students to retrieve and store resources of
interest, organize them in folders, bookmark, tag, rate, comment and share them with
peers. Students can form groups around various topics of interest, in which they collect
resources on a particular subject.
Automatic searching is also proposed in [1], by means of a federated search engine
which helps students retrieve resources from various services (MIT OCW or OpenER
courseware, Blogger, Technorati, YouTube, Slideshare etc.). Learners can organize
resources into collections, tag, rate and share them with peers. A filtering option is
provided, based on the popularity of the resources (in terms of the number of comments,
links, saves, likes, ratings, votes, views, shares, trackbacks etc.).
A similar approach is presented in [9]; Edu3R system allows students to search
through various learning object repositories (Ariadne, comPADRE, Connexions, LOR-
NET, Merlot, OCW, OER etc.) and save resources of interest. These can be subsequently
tagged, rated and shared with peers. A collaborative filtering mechanism is also included,
which recommends learning resources based on student similarity.
Curating Educational Resources for Homework Management 193
3 EdReHo Prototype
3.1 Concept, Features and Implementation
EdReHo is a web application designed to provide learning content curation support for
both teachers and students. The aim is to allow the collection and sharing of educational
resources needed to understand and solve homework assignments. When a teacher pro-
poses an assignment, he/she should be able to recommend also a set of resources relevant
for that topic. These resources are aimed to supplement the mandatory course material
and could take various forms, both formal and informal, such as: additional lecture slides,
developer guides, tutorials, video demonstrations, blog posts, source code examples on
GitHub, StackOverflow answers etc. In addition, students could also search for useful
resources on the web and share them with peers in case of group assignments. Peers’
recommendations have the potential to be very relevant, as they reflect the preference of
fellow students in a relatively homogenous learning community (e.g., a class centered
around the same course, in which students have similar learning backgrounds) [9].
In addition, the teacher can always access the resources added by the students to an
assignment and provide feedback if needed; in case of a useful and relevant resource,
the instructor can choose to save it to his/her own collection of resources and link
it to that particular assignment in the future. Conversely, in case of an incorrect or
irrelevant resource, the teacher can use the built-in comment feature to signal this issue
to the students. The possibility to add comments to assignments also encourages student
communication and interaction with peers and with the instructor.
Furthermore, EdReHo allows both students and teachers to tag and rate resources.
Tagging facilitates labeling and categorization of resources; subsequent retrieval is also
194 A.-I. Bala et al.
made easier by adding meaningful keywords. Tags also provide a personalized classifi-
cation, which is relevant to the learning community [5]. In addition, the quality of the
resources can be assessed both by students and instructors, by means of a simple 1 to 5
rating scale.
Some more details and illustrations of EdReHo functionalities are included in the
following subsection. As far as the implementation is concerned, EdReHo was developed
using mainly JavaScript as programming language. The following technologies were
used on the client side: Vue.js (an open source JavaScript framework for building the user
interface), Pug.js (a template engine used to inject data to produce HTML content) and
Stylus (a dynamic stylesheet preprocessor language providing an efficient and expressive
way to generate CSS). The server side is based on Node.js (an asynchronous event
driven JavaScript runtime environment designed to build scalable web applications) and
Express.js (the de facto standard server framework for Node.js), together with MongoDB
(a document-oriented database for storing all persistent data).
The resource management module offers support for learning content curation and
enriching the learning resources database. Thus, the instructor can save a resource of
interest by providing the link (URL) together with a title, a description, a set of tags and
a rating. Subsequently, the teacher can visualize the list of resources and search /filter
them (as displayed in Fig. 2).
The homework management module allows the instructor to create homework assign-
ments and link recommended resources to them, as shown in Fig. 3. The teacher can also
visualize the list of assignments, including the resources and comments added by the
students and communicate with the learners if needed. Finally, the instructor can also
access students’ solutions to the homework and evaluate them by providing a grade and
a feedback.
Curating Educational Resources for Homework Management 195
The student can also act as learning content curator in EdReHo. The system provides
the same functionalities with respect to resource management (adding, visualizing and
searching educational resources). As far as homework is concerned, the student can
visualize the list of assignments and select the one she/he prefers to solve. Subsequently,
the learner can add some relevant resources for the assignment (in addition to the ones
recommended by the teacher) and share them with peers; comments can also be posted
to an assignment, ensuring communication with the teacher and fellow students (as
illustrated in Fig. 4).
Finally, the student can also submit a solution for an assignment (and upload a
corresponding file); once the solution is assessed by the teacher, the learner can visualize
196 A.-I. Bala et al.
the feedback and grade. In addition, the system also sends email notifications to the
learners when their solutions are evaluated by the instructor.
4 Conclusion
We designed and implemented EdReHo, a support prototype for curating educational
resources, tailored to homework management. The system allows both teachers and
students to become content curators, collecting and sharing educational resources rele-
vant for a homework assignment. The process of searching, filtering, rating and tagging
resources has the potential to increase students’ engagement and critical thinking.
The next step is to experimentally evaluate the platform in various course settings.
In addition, we plan to extend EdReHo with a recommender module, which can suggest
resources of interest based on student’s profile. A guided tagging approach could also
be included, by automatically proposing keywords based on the resource content.
References
1. Chatti, M.A., Anggraeni, Jarke, M., Specht, M., Maillet, K.: PLEM: a Web 2.0 driven long
tail aggregator and filter for e-learning. Int. J. Web Inf. Syst. 6(1), 5–23 (2010)
2. Dettori, G., Torsani, S.: An approach to exploit social bookmarking to improve formal lan-
guage learning. In: Popescu, E., Li, Q., Klamma, R., Leung, H., Specht, M. (eds.) ICWL
2012. LNCS, vol. 7558, pp. 1–10. Springer, Heidelberg (2012). https://doi.org/10.1007/978-
3-642-33642-3_1
Curating Educational Resources for Homework Management 197
3. Gutierrez, K.: Why You Should Use Content Curation in Your L&D Strategy? (And How to Do
It) (2018). https://www.shiftelearning.com/blog/content-curation-learning-and-development
4. Jansen, B.J., Booth, D., Smith, B.: Using the taxonomy of cognitive learning to model online
searching. Inf. Process. Manage. 45, 643–663 (2009)
5. Kalamatianos, A., Zervas, P., Sampson, D.: ASK–LOST 2.0: a web-based tool for social
tagging of digital educational resources. In: Proceedings of 9th IEEE International Conference
on Advanced Learning Technologies (ICALT 2009), pp. 157–159 (2009)
6. Kiu, C., Lim, E.: Social bookmarking systems to enhance students’ learning process. In:
Proceedings of 3rd International Conference on Science in Information Technology (ICSITech
2017), pp. 413–417 (2017)
7. Marenzi, I., Zerr, S.: Multiliteracies and active learning in CLIL - the development of
LearnWeb2.0. IEEE Trans. Learn. Technol. 5(4), 336–348 (2012)
8. Pink, A.: Content Curation for Learning. https://anderspink.com/documents/content-
curation-book.pdf (2017)
9. Popescu, E., Buse, F.E.: Supporting students to find relevant learning resources through social
bookmarking and recommendations. In: Proceedings of 18th International Conference on
System Theory, Control and Computing (ICSTCC 2014), pp. 458–463 (2014)
10. Verhaart, M.: Curating digital content in teaching and learning using wiki technology. In:
Proceedings of 12th IEEE International Conference on Advanced Learning Technologies
(ICALT 2012), pp. 191–193 (2012)
11. Zhang, X.: Searching interactions and perceived learning. In: Zaphiris, P., Ioannou, A. (eds.)
LCT 2016. LNCS, vol. 9753, pp. 245–255. Springer, Cham (2016). https://doi.org/10.1007/
978-3-319-39483-1_23
UMLL (User Modeling and Language
Learning)
The Analysis of Worldwide Research
on Artificial Intelligence Assisted User Modeling
Xieling Chen1 , Dongfa Gao2 , Yonghui Lun3 , Dingli Zhou4 , Tianyong Hao4(B) ,
and Haoran Xie5(B)
1 Department of Mathematics and Information Technology,
The Education University of Hong Kong, Hong Kong SAR, China
s1131872@s.eduhk.hk
2 School of Information Science and Technology, Guangdong University of Foreign Studies,
Guangzhou, China
gaodf@gdufs.edu.cn
3 Guangzhou Huagong Information Software Co., LTD., Guangzhou, China
wingfai_lun@163.com
4 School of Computer Science, South China Normal University, Guangzhou, China
zhoudingli1999@126.com, haoty@m.scnu.edu.cn
5 Department of Computing and Decision Sciences, Lingnan University,
Hong Kong SAR, China
hrxie2@gmail.com
1 Introduction
and systems for interaction purposes, for example, intelligent tutoring systems,
computer aided assessment, and computer mediated communication [2]. With the abil-
ity of providing rich features about users, user modeling is of great significance for the
development of interactive systems as to offer various adaption effects [3, 4]. Different
types of user activities can be recorded and used to identify regularities in user paths
with the use of data mining and machine learning, which can be further integrated to
implicitly and dynamically generate user models [3]. For example, classification tech-
niques can map user information into different groups for the representation of various
user profiles [5].
Emerging interdisciplinary research towards AI assisted user modeling has attracted
increasing interests from the academia with a growing research articles available. It is of
need to handle the information to uncover important issues. Although there are reviews
concerning the adoption of data mining to user modeling within web-based personalized
educational systems [6–8], none of them is conducted by the use of quantitative approach.
And further, important issues, e.g., how do the research studies distribute by year, who
are the most active researchers and institutions, and what are the foci and hotspots related
to the field, have not been uncovered.
As a commonly adopted technique in the field of library and information sciences,
bibliometric analysis has long been regarded as an effective method for mapping scien-
tific articles in relation to a specific area [9]. It explores the distribution of articles based
on given categories such as topic, research subject, author, or country using quantitative
and statistical approaches. Bibliometric analysis have been widely used to investigate
research trend of a specific field recently [10–19].
This study aims to bibliometrically evaluate, academic articles about AI assisted user
modeling research published in Web of Science between 2001 and 2018. These articles
are analyzed and evaluated based on a number of perspectives (publication year, research
subjects, journals, authors, institutions, and countries/regions, as well as keywords) and
are employed to uncover major research issues and trends in the area.
ISI Web of Science (WOS) database was utilized to collect research articles concerning
AI assisted user modeling, since it is the most authoritative academic publication and cita-
tion repository. A retrieval field Topic (TS) in WOS, referring to title, abstract and/or key-
words, was used. The article retrieval was conducted in March 25, 2019. To acquire rele-
vant academic articles, an essential step was to prepare keywords lists for both AI and user
modeling research in education area. The identified keywords were then used to retrieve
articles from the database. Referring to the work of Hassan et al. [20], we followed the fol-
lowing steps to obtain relevant keywords for both AI and user modeling research. Step 1: A
list of seed keywords closely related to user modeling research in education area was pro-
vided by domain experts. Step 2: A query containing the seed keywords was constructed
with a restriction of Web of Science Category concerning ‘education’ or ‘educational’ to
retrieve articles with keywords matched against title, abstracts, and author defined key-
words. A part of the query to retrieve articles related to user modeling research in educa-
tion area was (‘user model*’ OR ‘user interface model*’ OR ‘learner model*’ OR ‘student
The Analysis of Worldwide Research on Artificial Intelligence 203
1 https://wordart.com/.
204 X. Chen et al.
The WoS subject category taxonomy is utilized for the analysis of research subject
distribution. The 333 articles are distributed in 22 subjects in total . Figure 2 shows the
The Analysis of Worldwide Research on Artificial Intelligence 205
top 10 subjects of articles, Education & Educational Research is the dominate subject
since articles retrieved are restricted to education relevant domain. It is worth noting
that Computer Science, Interdisciplinary Applications is ranked at 2nd . The two subjects
contribute 69.60% in total to the research field.
The 333 articles are identified to be published in 75 journals. Table 1 displays the
top 15 journals by H-index. They together account for 31.23% of the total articles.
Computers & Education is the dominate journal with an H-index value up to 30 and an
article count of 71. Other top journals include Educational Technology & Society and
IEEE Transactions on Learning Technologies. Further referring to Fig. 3 on both article
count and H-index, the dominate role of Computers & Education is significant.
398 institutions have participated in the research work, yet 76% of them have only
one research article. Table 3 displays the top 16 institutions ranked by H-index, among
which 9 are from Taiwan, again demonstrating a dominate position and contribution of
Taiwan. National Cheng Kung University and National Taiwan University of Science
and Technology are top two institutions.
National Council for Science and Technology and Chung Hua University have high
ACP values (78.00 and 78.25) although with relatively less articles (4 articles each),
indicating the high quality of their research articles. For most institutions, the interna-
tional collaboration rates are higher than 50%, especially National Council for Science
and Technology, Athabasca University, and Chung Hua University with an international
collaboration rate of 100%. The ACP values of internationally collaborated articles for
most institutions are higher than that of articles without international collaboration.
936 authors have participated in the research articles, yet 91% of them have only one
research article in the dataset. Table 4 displays the top 18 authors ranked by H-index,
among which 9 are from Taiwan. Gwo-Jen Hwang and Chih-Ming Chen are top two
208 X. Chen et al.
authors. It is worth noting that Patricio Garcia and Silvia Schiaffino have high ACP val-
ues (97.00 each) although with relatively less articles (3 articles each), indicating the high
quality of their research articles.
Top frequently used terms in different time periods are identified, as shown in Table 5.
For the period 2001–2018, the top terms ranked by frequency include ‘environment’
(44.74%), ‘adaptive’ (41.14%), and ‘intelligent’ (40.84%). From the trend test results
in the table, terms such as ‘adaptive’ ‘performance’, and ‘technology’ have experienced
a significant growth over the study periods, while terms ‘web’ and ‘information’ have
experienced a significant decreasing trend.
Figure 4 depicts the major terms covered in the articles in each time period. For
period 2001–2006, less terms are involved due to limited articles available. Top impor-
tant terms include ‘intelligence’, ‘tutoring’, ‘web’, and ‘learner’. For period 2007–2012,
top important terms include ‘learner’, ‘intelligence’, ‘tutoring’, and ‘environment’. For
period 2013–2018, top important terms include ‘environment’, ‘adaptive’, ‘education’,
‘data’, ‘performance’, and ‘self’. By comparing the three periods, some interesting find-
ings are as follows. Firstly, terms such as ‘education’ and ‘educational’ are gaining
constant attention, indicating that they are always main foci in the research. Secondly,
terms such as ‘adaptive’, ‘environment’, ‘data’, ‘self’, and ‘performance’, are becoming
more and more important. This indicates that there are growing interests among scholars
210 X. Chen et al.
in user adaptive learning research. Further, terms such as ‘intelligence’ and ‘tutoring’
are getting less and less attention with time going on. In addition, some terms have
experienced sudden change of attention. For example, term ‘web’ enjoys great attention
in the first period, however, it disappears suddenly in the latter two periods. Comparing
with the first and last periods, term ‘learner’ gets the most importance in the second
period.
4 Conclusion
This study presents the analysis of articles in the research field of AI assisted user model-
ing during years 2001–2018 from a bibliometric point of view. Based on the 333 retrieved
articles, this paper recognizes influential authors, institutions, countries/regions, as well
as journals and subjects, and reveals distribution and evolution of topics. Trend analysis
of article count indicates a continuing development of the research field. The analyzing
results can potentially benefit scholars in the field by raising their aware of the research
status and research topic evolution. Admittedly, in this study, only WoS was adopted to
retrieve research articles. Thus, further investigations would be needed to include more
relevant articles indexed by other databases like Scopus.
Acknowledgements. This work was supported by National Natural Science Foundation of China
(No.61772146).
References
1. Mwanza, D., Engeström, Y.: Pedagogical Adeptness in the Design of E-Learning Envi-
ronments: Experiences from the Lab@ Future project. In: E-Learn: World Conference on
E-Learning in Corporate, Government, Healthcare, and Higher Education, pp. 1344–1347.
AACE (2003)
212 X. Chen et al.
2. Leow, F.T., Neo, M.: Interactive multimedia learning: innovating classroom education in a
Malaysian University. Turkish Online J. Educ. Technol.-TOJET 13(2), 99–110 (2014)
3. Papatheocharous, E., Belk, M., Germanakos, P., Samaras, G.: Towards implicit user modeling
based on artificial intelligence, cognitive styles and web interaction data. Int. J. Artif. Intell.
Tools 23(02), 1440009 (2014)
4. Brusilovsky, P., Millán, E.: User models for adaptive hypermedia and adaptive educational
systems. In: Brusilovsky, P., Kobsa, A., Nejdl, W. (eds.) The Adaptive Web. LNCS, vol. 4321,
pp. 3–53. Springer, Heidelberg (2007). https://doi.org/10.1007/978-3-540-72079-9_1
5. Wu, X., Kumar, V., Quinlan, J.R., Ghosh, J., Yang, Q., Motoda, H., et al.: Top 10 algorithms
in data mining. Knowl. Inf. Syst. 14(1), 1–37 (2008)
6. Mobasher, B.: Data mining for web personalization. In: Brusilovsky, P., Kobsa, A., Nejdl, W.
(eds.) The Adaptive Web. LNCS, vol. 4321, pp. 90–135. Springer, Heidelberg (2007). https://
doi.org/10.1007/978-3-540-72079-9_3
7. Pierrakos, D., Paliouras, G., Papatheodorou, C., Spyropoulos, C.D.: Web usage mining as a
tool for personalization: a survey. User Model. User-Adap. Inter. 13(4), 311–372 (2003)
8. Eirinaki, M., Vazirgiannis, M.: Web mining for web personalization. ACM Trans. Internet
Technol. (TOIT) 3(1), 1–27 (2003)
9. Pérez-Acebo, H., Linares-Unamunzaga, A., Abejón, R., Rojí, E.: Research trends in pavement
management during the first years of the 21st century: a bibliometric analysis during the
2000–2013 period. Appl. Sci. 8(7), 1041 (2018)
10. Song, Y., Chen, X., Hao, T., Liu, Z., Lan, Z.: Exploring two decades of research on classroom
dialogue by using bibliometric analysis. Comput. Educ. 137, 12–31 (2019)
11. Chen, X., Wang, S., Tang, Y., Hao, T.: A bibliometric analysis of event detection in social
media. Online Inf. Rev. 43(1), 29–52 (2019)
12. Hao, T., Chen, X., Li, G., Yan, J.: A bibliometric analysis of text mining in medical research.
Soft. Comput. 22(23), 7875–7892 (2018)
13. Chen, X., Liu, Z., Wei, L., Yan, J., Hao, T., Ding, R.: A comparative quantitative study of
utilizing artificial intelligence on electronic health records in the USA and China during
2008–2017. BMC Med. Inform. Decis. Mak. 18(5), 117 (2018)
14. Chen, X., Lun, Y., Yan, J., Hao, T., Weng, H.: Discovering thematic change and evolution
of utilizing social media for healthcare research. BMC Med. Inform. Decis. Mak. 19(2), 50
(2019)
15. Chen, X., Ding, R., Xu, K., Wang, S., Hao, T., Zhou, Y.: A bibliometric review of natural
language processing empowered mobile computing. Wireless Communications and Mobile
Computing, 2018 (2018)
16. Chen, X., Xie, H., Wang, F.L., Liu, Z., Xu, J., Hao, T.: A bibliometric analysis of natural
language processing in medical research. BMC Med. Inform. Decis. Mak. 18(1), 14 (2018)
17. Chen, X., Hao, J., Chen, J., Hua, S., Hao, T.: A bibliometric analysis of the research status of
the technology enhanced language learning. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau,
R. (eds.) SETE 2018. LNCS, vol. 11284, pp. 169–179. Springer, Cham (2018). https://doi.
org/10.1007/978-3-030-03580-8_18
18. Chen, X., Weng, H., Hao, T.: A data-driven approach for discovering the recent research status
of diabetes in China. In: Siuly, S., et al. (eds.) HIS 2017. LNCS, vol. 10594, pp. 89–101.
Springer, Cham (2017). https://doi.org/10.1007/978-3-319-69182-4_10
19. Chen, X., Chen, B., Zhang, C., Hao, T.: discovering the recent research in natural language
processing field based on a statistical approach. In: Huang, T.-C., Lau, R., Huang, Y.-M.,
Spaniol, M., Yuen, C.-H. (eds.) SETE 2017. LNCS, vol. 10676, pp. 507–517. Springer, Cham
(2017). https://doi.org/10.1007/978-3-319-71084-6_60
20. Hassan, S.U., Haddawy, P., Zhu, J.: A bibliometric study of the world’s research activity in
sustainable development and its sub-areas using scientific literature. Scientometrics 99(2),
549–579 (2014)
The Analysis of Worldwide Research on Artificial Intelligence 213
21. Zawacki-Richter, O., Latchem, C.: Exploring four decades of research in computers &
education. Comput. Educ. 122, 136–152 (2018)
22. Zhong, S., Geng, Y., Liu, W., Gao, C., Chen, W.: A bibliometric review on natural resource
accounting during 1995–2014. J. Cleaner Prod. 139, 122–132 (2016)
23. Gimenez, E., Salinas, M., Manzano-Agugliaro, F.: Worldwide research on plant defense
against biotic stresses as improvement for sustainable agriculture. Sustainability 10(2), 391
(2018)
24. Hirsch, J.E., Buela-Casal, G.: The meaning of the H-Index. Int. J. Clin. Health Psychol. 14(2),
161–164 (2014)
25. Mann, H.B.: Nonparametric tests against trend. Econometrica J. Econometric Soc. 13(3),
245–259 (1945)
Analyzing the Influence of Academic
Papers Based on Improved PageRank
1 Introduction
The academic influence has always been a high concern of research scholars, and
it plays a decisive role in fund application, talent plan declaration and so on.
The influence of papers is an important part of academic influence. High-impact
papers often give authors a better reputation and inspire others, even lead to a
new field. The importance of high-impact papers is obvious, so the key question
is how to judge whether these are high-impact papers, therefore, the evaluation
method is particularly important. In general, two methods are commonly used
to evaluate the value of the impact of the paper. One is based on the Impact
Factor (IF) of journals [1], the other one is based on the citation number [2,3].
A high-IF journal not always publishes high-impact papers, thus, the journal
IF is not suitable for the assessment of the impact of papers [4]. The key point of
journal IF is based on citation number. The more citation journals obtain, the
higher IF journals have. So without the number of citations, the IF of journals
makes no sense. The number of times the papers were cited can reflect the papers’
influence under some certain conditions, but there are still some issues.
The first issue is new papers are less cited than the old, old papers have longer
exposure time, so the citation numbers are bigger than new papers generally. This
will lead to a special situation, an old paper is considered to be more influential
than a new paper by the method of citation number, but this is not true in reality.
Second issue is the different periods of citation have different meanings, it has
been claimed in [5] that short-term citations can be considered as currency at
the research front, whereas long-term citations can contribute to the codification
of knowledge claims into concept symbols. In order to solve these problems, this
article studies some related content.
We mentioned two issues above. For the first issue, this article proposes a
method that combines the authors’ impact. Influence of authors is an important
indicator of the paper’s influence. For those authors who have published many
high-impact papers, their next new paper may also be high-impact too. So we
can roughly infer the impact value of the author’s new papers when we get the
impact value of the author. For the second issue, this article suggests a method
that considers the year interval of two papers (paper and its reference). For
the evaluation of papers, short-term citations are not significant as long-term
citations, so we can combine the year interval to weight the impact value that
reflects the different citations.
The contribution of this paper is that we adopt the GapYear-Rank approach
to improve the existing evaluation methods we have discussed, combining the
authors’ influence and the structure in the citation network.
2 Related Work
The indicators of papers influence and calculation methods are attracting argues
for a long time, we discuss these related three issues. They are the relationship
between citation and papers’ influence, the difference between short-term and
long-term citations, and the method of calculating influence. We will discuss
these issues in the rest.
According to research by Filippo Radicchi et al. [6], when scientists have
enough information and can indeed make objective and fair choices, their judg-
ment on the impact of papers is consistent with the number of papers been
cited. That is to say, papers with more citation have greater influence. A paper,
although released in a high IF publication for a few years, and it is rarely quoted,
indicating that whether it is released or not, except for the author himself, it’s
the same for peers. In other words, this paper has little influence on the develop-
ment of the subject area [7]. So a high-impact paper usually has more citation,
evaluating papers according to the number of citation is a proper way to work.
There is a difference between short-term citation and long-term citation, the
summary of [5] shows that a majority of the indicators used for the evaluation
for the papers and journals are biased towards short-term impact, it can be
expected to lead to a selection bias that is skewing the results of evaluations in
favor of short-term impact. It may lead scholars to prefer to publish papers on
popular topics for higher impact factor to improve the reputation and influence
216 C. Ji et al.
of himself while ignoring some other important topics and underestimating some
scholar who made real contributions in some unpopular field. There is a special
case, called sleeping beauties [8]. Sleeping beauty in science refers to a paper
whose importance is not recognized for decades after publication. Its citation
history exhibits a long hibernation period followed by a sudden spike of pop-
ularity. Considering these conditions and this issue, this article separates these
conditions and carries out different processing to make the evaluation of papers
more objective and fair.
PageRank algorithm is one of the methods that Google used to evaluate
the webpages’ impact to improve the quality of web search engines [9]. It helps
Google a lot in the search engine field and dictates the rules for everyone else [10].
This algorithm has been widely applied not only to rank web search results but
also to rank the academic papers [11]. PageRank is a mathematical algorithm
that evaluates the quality and quantity of links to a webpage. This evaluation
helps it to determine a relative score of the page’s importance and authority
[12]. This algorithm draws on the general method of evaluating the importance
of papers in academia, by using the data about citation to evaluate the web-page,
so using PageRank algorithm to evaluate the influence or quality of papers is a
backtracking practice and suitable method.
The basic idea of PageRank is based on two assumptions. One is the quantity
assumption, if page x is linked by more pages, it means that the more important
x is. Another is the quality assumption, if the page y is linked by a higher
quality page, it means that the quality of y is higher. Intuitively, qq.com created
by Tencent is a famous, popular, high-impact page in China, reflected by the
fact that many pages link it. Likewise, pages prominently pointed to from qq.com
are themselves probably important [13]. Let PR(x) represent the importance of
page x, so we can use the out-links of page x to calculate the PR(x). Generally
speaking, we pre-give an initial PR for each web page as 1, and N is the total
number of web’s out-link. For example, Nx is the out-degree of page x. since
the physical meaning of the PR is the access probability of a web page. Let By
represent the set of pages pointing to y. In each iteration, propagate the ranks
as follows:
P R(x)
P R(y) = (1)
Nx
x∈By
But there are two special situations should be concerned. One is that some
selfish web-pages which do not have any out-link except itself, its value of PR
will only increase while iterating. Another is web-pages do not have out-link
including itself, all the PR will become zero while iterating. Both of them are
unreasonable. To solve these special situations, we can image real people who
surf on the internet, while facing these web-pages, they won’t be trapped here.
We assume that they have a certain probability of inputting URL to jump to a
random web-page directly, and the probability of jumping to each web-page is
the same, and we use α to represent the probability. The improved formula is as
follows:
Analyzing the Infuence of Academic Papers Based on Improved PageRank 217
1−α P R(x)
P R(y) = +α (2)
M Nx
x∈By
If the page i doesn’t link the page j, the l(pi , pj ) = 0 for each j. the matrix
should satisfy one condition:
M
l (pi , pj ) = 1 (4)
i=1
3 The Method
3.1 Overview of Proposed Evaluation Methods
Our proposed evaluation method combines the influence of the authors, this
method considers the relationship among the papers in the citation network
and evaluates the influence of authors by their papers. The process for the pro-
posed method comprises five steps: (1) cleaning dataset and structuring it into
database, (2) calculating the initial papers’ PR in citation network by using
PageRank, (3) calculating the authors’ impact value by initial papers’ PR, (4)
calculating the initial papers’ PR from authors’ impact value and (5) calculating
the papers’ final PR by using improved PageRank named GapYear-Rank. An
overview of the procedure for the proposed method is shown in Fig. 1.
main point is the more you contribute, the more points you add. The first author
can get all the impact value of the paper, the second author can get half of that,
we especially take the corresponding author into consideration, and treat them
as the second author. These formulas are as follows:
We use InitPR(x) to represent the initial value of the paper based on the authors’
impact, and add all the processed author(z) from paper x to get all the InitPR(x).
This operation solves the problem of cold start in PageRank and makes the
result of evaluation more accurate and reasonable. But it is not enough to use
the variable of author impact to evaluate the paper, we still need a citation
network in the next work.
3.4 GapYear-Rank
And we initiated the first iteration matrix of papers’ PR based on the authors’
impact, high impact author has a larger base in the GapYear-Rank algorithm’s
iterations, this can solve the issue of new paper has a smaller cited number. The
weight in GapYear-Rank algorithm can solve the issue of citation in a different
term.
AMiner is designed to search and perform data mining operations against aca-
demic publications on the Internet, using social network analysis to identify con-
nections between researchers, conferences, and publications [14], it also provides
some open datasets. Our dataset comes from the Citation Network Dataset of
AMiner [15]. We choose the smaller dataset that contains 629,814 blocks. Each
block contains the following information: (1) Index: every paper has unique index
220 C. Ji et al.
value, (2) Title: paper’s title, (3) Year: the published year of paper, (4) Author:
authors who participated in the publishing of this paper, (5) Reference: the
index of references of this paper (there are multiple lines, with each indicating
a reference). Table 1 is a basic outline of the data obtained.
Paper is every individual paper, Citation contains the paper’s index and
reference’s index, Author is the authors of each paper are stored separately.
Based on this dataset, we write three tables into a MySql database and generate
two tables in the process of calculating PR.
Table 4. Top 10 ordered by papers’ impact based on authors’ impact and GapYear-
Rank.
Fig. 2. The PR score of all papers based on PageRank algorithm and the author’s
impact with GapYear-Rank algorithm.
We find that the citation number of the third paper is 50 times more than the
second paper from the Table 3. That is because the nine papers which cited the
second paper are all high quality, their average value of initial PR has reached
1.90513e-05, at the same time, the average value of third paper is 7.69379e-07, so
the second paper entered the top 10 by high quality, and the third paper entered
the top 10 by huge quantity.
From Table 4, we will find that there are eight papers in both leaderboards
while comparing with Table 2, and five papers while comparing with Table 3. It
shows that the results of the three methods in this paper are generally close.
Table 5 shows the top 10 influential authors through our calculation based on
the initial PR score of papers, author Charles T. Davies and author Lawrence
A. Bjork are published only one paper but entered the top 10, the reason is that
the method which we choose the top 10 authors is based on the average score of
authors’ published papers’ initial PR, author Charles T. Davies, and Lawrence
A. Bjork are both published a high impact value paper, so they could get high
author’s impact by their paper.
The results of Tables 2, 3, and 4 are generally similar. But among them, the
differences in Table 3 are more obvious, the citation number in the first and
second rows of the table is less than other rows. In order to more intuitively
refer to all data, we have made a distribution of the different number of the
top papers based on these methods in Fig. 3. It is not difficult to find that the
citation number line differs greatly from the other two lines, and the citation
number line is roughly at the center of the two lines. From the variance analysis
of the citation number of different years in Table 6, we find that the result of
the citation number is larger than other methods in each column. It means
the result of citation number has a large fluctuation in different years, which
is unreasonable because high-impact papers always appear in a period of time.
To some extent, the smaller variance results are fairer. Therefore, according to
Fig. 3 and Table 6, PageRank and GapYear-Rank performed better than the
citation number. That is because PageRank and GapYear-Rank also consider
the relationship between citation networks, that is to say, these two methods
not only take the quantity but also the quality into consideration.
But even the variance results of PageRank is smaller than GapYear-Rank, we
can’t conclude that the PageRank is better than GapYear-Rank, cause they both
consider the citation numbers and citation relationships. So we are supposed to
analyze deeper data.
We will find different PR score distributions from the two subgraphs in
Fig. 2, we take some examples to illustrate the differences. The 9th top paper
in GapYear-Rank, which is ranked 29th in PageRank has 553 cited papers, the
cited number in ten years only accounted for 164, about 29.66% of the total, but
in 2008 and 2009, the cited number has reached for 235, about 43.50% of the
total. So this paper is probably the sleeping beauty we mentioned above, similar
situations have appeared in other papers, such as the 8th and 10th top paper
in GapYear-Rank, both of them are sleeping beauties and the 8th paper which
titled C4.5: programs for machine learning has a great influence on the field of
machine learning. From these results, we find that GapYear-Rank is better to
find the influential papers and rank them in higher rankings.
Fig. 3. The distribution of the different number of top papers based on these methods.
5 Conclusion
References
1. Du, M., Bai, F., Liu, Y.: PaperRank: a ranking model for scientific publications.
In: 2009 WRI World Congress on Computer Science & Information Engineering,
vol, 4, pp. 277-281 (2009). https://doi.org/10.1109/CSIE.2009.479
2. Beel, J., Gipp, B.: Google scholar’s ranking algorithm: the impact of articles’
age (an empirical study). In: 2009 Sixth International Conference on Informa-
tion Technology: New Generations, pp. 160–164 (2009). https://doi.org/10.1109/
ITNG.2009.317
3. Beel, J., Gipp, B.: Google scholar’s ranking algorithm: the impact of citation counts
(an empirical study). In: 2009 Third International Conference on Research Chal-
lenges in Information Science, pp. 439–446 (2009). https://doi.org/10.1109/RCIS.
2009.5089308
4. Mariethoz, G., Karssenberg, D., Grana, D.: Who cares about impact factor?. J.
Comput. Geosci. 115, iii–iv (2018)
5. Leydesdorff, L., Bornmann, L., Comins, J.A., Milojevic, S.: Citations: indicators
of quality? The impact fallacy. J. Front. Res. Metrics Analytics 1, 1 (2016)
6. Radicchi, F., Weissman, A., Bollen, J.: Quantifying perceived impact of scientific
publications. J. Informetrics. 11(3), 704–712 (2017)
7. Wen, S.: Papers with more citations may not be influential. News. China Science
Daily. 2 (2017)
8. Ke, Q., Ferrara, E., Radicchi, F., et al.: Defining and identifying Sleeping Beauties
in science. Proc. Nat. Acad. Sci. U.S.A. 112(24), 7426 (2015)
9. Liang, Y., Li, Q., Qian, T.: Finding relevant papers based on citation relations. In:
Web-age Information Management, pp. 403–414 (2011). https://doi.org/10.1007/
978-3-642-23535135
10. SEO PowerSuite: Beginner’s Guide to Google PageRank: How It Works & Why It
Still Matters in 2018. https://www.link-assistant.com/news/page-rank-2018.html.
Accessed 20 Jul 2019
11. Gipp, B., Beel, J., Hentschel, C.: Scienstein: a research paper recommender system.
In: Proceedings of the International Conference on Emerging Trends in Computing
(ICETIC 2009), pp. 309–315 (2009)
12. Son, J.: Seoung Bum Kim: academic paper recommender system using multilevel
simultaneous citation networks. J. Decis. Support Syst. 105, 24–33 (2018)
13. Haveliwala, T.H: Topic-sensitive PageRank. In: International Conference on World
Wide Web, pp. 517–526 (2002). https://doi.org/10.1145/511446.511513
14. Tang, J., Zhang, J., Yao, L., et al.: ArnetMiner: extraction and mining of academic
social networks. In: Proceedings of the ACM SIGKDD International Conference
on Knowledge Discovery and Data Mining, ACM, pp. 990–998 (2008). https://doi.
org/10.1145/1401890.1402008
15. AMiner: Citation Network Dataset. https://www.aminer.cn/citation. Accessed 20
Jul 2019
A Systematic Review of Frameworks for Coding
Towards Classroom Dialogue
haoty@m.scnu.edu.cn
1 Introduction
believe that classroom dialogue is beneficial in terms of fostering thinking and learning,
and particularly facilitates deep processing of knowledge (Howe et al. 2019). Making
effective use of dialogue to increase students’ learning achievement has been the main
aim of research. A definition proposed by Howe and Abedin (2013) referring to class-
room dialogue as the kind of communication in which one individual addresses another
individual or individuals and at least one addressed individual replies, is broad enough
to encapsulate many of these commonalities.
The number of publications on classroom dialogue has increased dramatically during
the past 20 years, and interests in research on, and practical applications of, classroom
dialogue has grown around the world (Song et al. 2019). When considering dialogue,
scholars usually use the codes that emerge from specific scripts or they design a frame-
work that is specific to their special research interests. Various coding frameworks,
instruments and specific codes have been developed, which has made cross-study com-
parisons difficult, and has prevented the field from exerting the greater influence on policy
and practice (Hennessy et al. 2016). Therefore a systematic review of the coding frame-
works used in the examination of classroom dialogue is greatly needed in order to allow
the emergence of typical themes, forms and functions across the frameworks. Scholars
and practitioners would become more competent in designing or selecting frameworks.
2 Data Selection
This sample selection was based on a previous work of the authors (Song et al. 2019).
‘Classroom’ and its two associated terms ‘whole-class’ and ‘small-group’ were taken as
one set of keywords and ‘dialogue’ and associated terms, ‘dialogic’, ‘discourse’, ‘conver-
sation’, ‘discussion’, ‘language’, ‘interaction’, ‘talk’, ‘communication’ and ‘speaking’
acted as the other set of keywords in literature search. Publications were retrieved from
the most important bibliometric database, Web of Science (WoS), and 3,914 publications
were filtered after computer retrieval and artificial preprocessing. We checked these pub-
lications and retained the ones showing complete or parts of coding frameworks, which
resulted in 51 articles remaining for reviewing. The follows information was systemati-
cally reviewed: name of instruments, contexts of the use of instruments (i.e. educational
contexts and cultural contexts), coding domains, specific codes, and evaluation of frame-
works (i.e. reliability and validity). A summary of the main coding domains is listed in
Fig. 1.
The third theme shown in coding frameworks is concerned with analysis, which
can further be found in evaluation, explanation/justification and exemplification (see
e.g. Boyd and Markarian 2011; Dolin et al. 2018; Hennessy et al. 2016; Nystrand et al.
2003). Evaluation is used to request or provide an opinion about how useful or appropri-
ate another person’s contribution (in words or action) has been; explanation/justification
refers to an argument to support a preceding contribution or to show that it was reason-
able or appropriate; exemplification requests more information, including examples, to
supplement existing, relatively simple information or to illustrate a point that has just
been made. Dialogue of this kind is characterized by the use of deductive reasoning
and making knowledge explicit and easy to understand. The setting of codes in this
theme aims to support students in digging deeper into their ideas and reasoning, and in
abstractly separating a whole into its constituent parts in order to study the parts and
their relationships.
The fourth theme is that coding should show the coordination of information, and
summarization, comparison and connection are its main specific embodiments (see
Alexander 2008; Kumpulainen and Lipponen 2014; Mercer and Littleton 2007; Reznit-
skaya and Gregory 2013). Summarization is a condensed statement of what has been
said by one or more participants or the provision of only the main information; compar-
ison refers to dialogue that examines the similarities or differences in knowledge and
information; connection assesses whether dialogue builds and explores the relationships
between things and information. This kind of dialogue helps students focus, connect,
coordinate and reflect, and promotes deep learning. It is a process of formulating gen-
eral concepts by reasoning from detailed facts and involves inductive reasoning and the
development of ideas.
The fifth theme concerns speculation, which refers to dialogue that considers pos-
sibilities, going beyond the current state of knowledge but with a theoretical or factual
basis (see e.g. Chin 2006; Mortimer and Scott 2003; Nystrand 1997; Ruiz-Primo 2011).
This kind of dialogue supports students’ generation of new ideas and develops creative
abilities.
The sixth theme is that categories should be able to assess how teachers and students
use uptakes (see e.g. Mortimer and Scott 2003; Ruiz-Primo 2011; Wells and Arauz
2006). The issue here concerns the way in which the previous response by somebody is
taken up by others. Uptake can be realized in a comment that explicitly recognizes the
previous response and builds on it in some way; or it can be manifested in a question that
asks the previous responder to elaborate on what s/he has said. Extension of talk provides
a constructive supplement based on previous talk, rather than simply a repetition of a
previous articulation (see e.g., Alexander 2008; Nystrand 1997; O’Connor et al. 2015;
Rojas-Drummond et al. 2010). The inclusion of this theme aims to encourage students
to listen carefully to their peers’ ideas; to build on, think and reason with others.
230 Y. Song et al.
Table 1. (continued)
People hold controversial views with regard to several main issues, which has led to the
variety of coding forms and functions. The issues are illustrated first, together with a
justification of our positionality.
purposefully selected excerpt and use codes that emerge from a transcript (Rampton
2009). This usually employs a linguistic ethnography and allows adequate contex-
tual detail and linguistic characteristics of dialogic teaching and learning to be
examined (Lefstein 2008). In contrast, coding frameworks based on the sociocul-
tural approach are more fundamental than those addressing language use and look
beyond the surface forms of dialogue to demonstrate the relationship between lan-
guage and thoughts (see e.g. Chin 2006; Fisher 2007; Muhonen et al. 2017). As a
learning method, the ultimate goal of involvement in classroom dialogue is that it
should prompt thinking and learning (Alexander 2017; Howe 2017). According to
the socio-cultural approach, knowledge is a socially mediated product and can be
built collaboratively between teacher and students or among students themselves
(Mercer and Dawes 2014). Coding frameworks should be able to account for how
dialogue acts as a social and cognitive mediation, and to elaborate the process of
narrating, expressing, analysing, generalizing, speculating and constructing knowl-
edge. From the review, more than 80% of publications have used the sociocultural
approach when assessing classroom dialogue. We agree with the view of the major-
ity that dialogue should go beyond its role as a linguistic tool in order to make
important contributions to improving the quality of the teaching–learning process.
This requires consideration of the cognitive function of language when developing
or selecting coding frameworks.
(2) Coding units and levels
The unit of coding and analysis varies across studies. Dialogue can be coded at
a single word, a sentence, or sentences in which at least one characterization or
function is clearly identifiable (Chin 2006; Wells and Arauz 2006). Some coding
frameworks focus on specific dialogic strategies or specific knowledge-building
techniques (e.g. Alexander 2017; Hennessy et al. 2016). Others are broader when
it comes to assessing dialogic activities or interactive behaviours during classroom
teaching and learning (e.g. Galton et al. 1999; Jay et al. 2017; Mortimer and Scott
2003). Levels of coding are also different, and coding can be applied at a macro
level or a micro level to elicit multi-dimensional information. Researchers, such as
Saville-Troike (2003) and Hennessey et al. (2016), have used an ethnography of
communication, which has established a system of hierarchical levels of analysis
consisting of ‘communicative acts’ (CA) at a micro level; these are embedded within
‘communicative events’ (CE) at a meso level; which are in turn part of broader ‘com-
municative situations’ (CS) at a macro level. Some others, for example, Wells and
Arauz (2006) and Nystrand et al. (2003) have proposed a five-level coding in which
from the macro to micro levels there are episodes, sequences, exchanges, moves
and acts. Howe et al. (2019) distinguish two levels of measurement of dialogue,
turn-level coding and lesson-level ratings.
The selection of unit and levels of coding depends on research interests and at the
same time, is influenced by methodology choice (Mercer 2010; O’Connor et al.
2015). A broad, macro-level coding is adaptable to large sample studies and quan-
titative analysis (Howe et al. 2019). It is time-consuming and costly to conduct data
collection and analysis when using a detailed and micro-level coding framework
(Gillies and Nichols 2015). Scholars tend to use qualitative research methods and
work with a small number of students and teachers, thus the results are less likely
A Systematic Review of Frameworks for Coding Towards Classroom Dialogue 233
5 Conclusion
Classroom dialogue, acting as a frequently used method for teaching and learning, has
attracted much interest around the world and publications in this field keep increas-
ing. The choice of coding frameworks matters in the assessment of dialogic teaching
and learning. Various coding frameworks, instruments and specific codes have been
developed, which has made cross-study comparisons difficult. This paper conducted
a systematic review of the coding frameworks used in the examination of classroom
dialogue. The review indicates that three issues are closely related to the development
or choice of frameworks: linguistic ethnography versus sociocultural approach, coding
units and levels, and objects of coding. At the same time, six main themes emerged
from many of these codes and these should be encapsulated in a coding framework,
which are prior knowledge, personal information, analysis, generalization, speculation
and uptakes. With this knowledge, scholars and practitioners would be more competent
in designing or selecting frameworks. Findings would be more feasible for comparisons
across different educational and cultural contexts, which will make it possible for the
field to exert greater influence on policy and practice.
References
Aguiar, O.G., Mortimer, E.F., Scott, P.: Learning from and responding to students’ questions: the
authoritative and dialogic tension. J. Res. Sci. Teach. 47(2), 174–193 (2010)
Alexander, R.J.: Essays on Pedagogy. Routledge, London (2008)
Alexander, R.J.: Teaching and learning for all? The quality imperative revisited. Int. J. Educ. Dev.
40, 250–258 (2015)
Alexander, R.J.: Developing dialogue: process, trial, outcomes. Paper of 17th Biennial EARLI
Conference, 31 August 2017, Tampere, Finland (2017)
Bakeman, R., Gottman, J.M.: Observing Interaction: An Introduction to Sequential Analysis.
Cambridge University Press, Cambridge (1997)
Boyd, M.P., Markarian, W.C.: Dialogic teaching: talk in service of a dialogic stance. Lang. Educ.
25(6), 516–534 (2011)
Chinn, C.A., O’Donnell, A.M., Jinks, T.S.: The structure of discourse in collaborative learning. J.
Exp. Educ. 69(1), 77–97 (2000)
Chin, C.: Classroom interaction in science: teacher questioning and feedback to students’
responses. Int. J. Sci. Educ. 28(11), 1315–1346 (2006)
Dolin, J., Bruun, J., Nielsen, S.S., Jensen, S.B., Nieminen, P.: The structured assessment dialogue.
In: Dolin, J., Evans, R. (eds.) Transforming Assessment. CSER, vol. 4, pp. 109–140. Springer,
Cham (2018). https://doi.org/10.1007/978-3-319-63248-3_5
Dwyer, J., Kelcey, B., Berebitsky, D., Carlisle, J.F.: A study of teachers’ discourse moves that
support text-based discussions. Elem. School J. 117(2), 285–309 (2016)
Ellegaard, M., Damsgaard, L., Bruun, J., Johannsen, B.F.: Patterns in the form of formative
feedback and student response. Assess. Eval. High. Educ. 43(5), 727–744 (2017)
Fisher, R.: Dialogic teaching: developing thinking and metacognition through philosophical
discussion. Early Child Dev. Care 177(6–7), 615–631 (2007)
Galton, M., Hargreaves, L., Comber, C., Wall, D., Pell, T.: Changes in patterns of teacher interaction
in primary classrooms: 1976-96. Br. Edu. Res. J. 25(1), 23–37 (1999)
Gillies, R.M., Nichols, K.: How to support primary teachers’ implementation of inquiry: teachers’
reflections on teaching cooperative inquiry-based science. Res. Sci. Educ. 45(2), 171–191
(2015)
Gillies, R.M., Baffour, B.: The effects of teacher-introduced multimodal representations and dis-
course on students’ task engagement and scientific language during cooperative, inquiry-based
science. Instr. Sci. 45(4), 493–513 (2017)
Harney, O., Hogan, M., Quinn, S., et al.: Investigating the effects of peer to peer prompts on col-
laborative argumentation, consensus and perceived efficacy in collaborative learning. Comput.
Support. Collab. Learn. 12(3), 307–336 (2017)
Heift, T.: Corrective feedback and learner uptake in call. Recall 16(1), 37–66 (2004)
Hennessy, S., et al.: Developing a coding scheme for analysing classroom dialogue across
educational contexts. Learn. Culture Soc. Interact. 9, 16–44 (2016)
Howe, C., Abedin, M.: Classroom dialogue: a systematic review across four decades of research.
Camb. J. Educ. 43(3), 325–356 (2013)
Howe, C.: Advances in research on classroom dialogue: commentary on the articles. Learn. Instr.
48, 61–65 (2017)
Howe, C., Hennessy, S., Mercer, N., Vrikki, M., Wheatley, L.: Teacher-student dialogue during
classroom teaching: does it really impact upon student outcomes? J. Learn. Sci. 28(4–5), 462–
512 (2019)
Hulsman, R.L., Vloodt, J.V.D.: Self-evaluation and peer-feedback of medical students’ commu-
nication skills using a web-based video annotation system. Exploring content and specificity.
Patient Educ. Couns. 98(3), 356–363 (2015)
A Systematic Review of Frameworks for Coding Towards Classroom Dialogue 235
Jadallah, M., et al.: Influence of a teacher’s scaffolding moves during child-led small-group
discussions. Am. Educ. Res. J. 48(1), 194–230 (2011)
Jay, T., et al.: Dialogic Teaching: Evaluation Report and Executive Summary. Education
Endowment, London (2017)
Jie, Z., Shahbaz, M., Chunling, L., Richard, C.A.: What makes a more proficient discussion group
in English language learners’ classrooms? Influence of teacher talk and student backgrounds.
Res. Teach. Engl. 51(2), 183–207 (2016)
Kumpulainen, K., Wray, D.: Classroom Interactions and Social Learning: From Theory to Practice.
Routledge, Abingdon (2003)
Kumpulainen, K., Lipponen, L.: Building on the positive in children’s lives: a co-participatory
study on the social construction of children’s sense of agency. Early Child Dev. Care 184(2),
211–229 (2014)
Lee, S.C., Irving, K.E.: Development of two-dimensional classroom discourse analysis tool
(CDAT): scientific reasoning and dialog patterns in the secondary science classes. Int. J. STEM
Educ. 5(1), 5 (2018)
Lefstein, A.: Changing classroom practice through the English National Literacy Strategy: a
micro-interactional perspective. Am. Educ. Res. J. 45(3), 701–737 (2008)
Lefstein, A., Snell, J., Israeli, M.: From moves to sequences: expanding the unit of analysis in the
study of classroom discourse. Br. Edu. Res. J. 41(5), 866–885 (2015)
Liu, Y.F., Zhao, Y.: A study of teacher talk in interactions in English classes. Chin. J. Appl.
Linguist. 33(2), 76–86 (2010)
Macgregor, S.K., Atkinson, T.R.: Facilitating learner interactions in the two-way video classroom.
J. Educ. Technol. Syst. 31(1), 45–62 (2002)
Mercer, N.: Words and Minds: How We Use Language to Think Together. Routledge, London
(2000)
Mercer, N., Dawes, L., Wegerif, R., Sams, C.: Reasoning as a scientist: ways of helping children
to use language to learn science. Br. Edu. Res. J. 30(3), 359–377 (2004)
Mercer, N., Littleton, K.: Dialogue and the Development of Children’s Thinking: A Sociocultural
Approach. Routledge, Abingdon (2007)
Mercer, N.: The analysis of classroom talk: methods and methodologies. Br. J. Educ. Psychol.
80(1), 1–14 (2010)
Mercer, N., Dawes, L.: The study of talk between teachers and students, from the 1970s until the
2010s. Oxf. Rev. Educ. 40(4), 430–445 (2014)
Mortimer, E., Scott, P.: Meaning Making in Secondary Science Classroomsaa. McGraw-Hill
Education, London (2003)
Muhonen, H., Rasku-Puttonen, H., Pakarinen, E., Poikkeus, A.M., Lerkkanen, M.K.: Knowledge-
building patterns in educational dialogue. Int. J. Educ. Res. 81, 25–37 (2017)
Nassaji, H., Wells, G.: What’s the use of “triadic dialogue”? An investigation of teacher-student
interaction. Appl. Linguist. 21(3), 376–406 (2000)
National Institute of Child Health and Human Development Early Child Care Research Network:
The relation of global first-grade classroom environment to structural classroom features and
teacher and student behaviors. Elem. School J. 102(5), 367–387 (2002)
Nichd, E., Bonnie, K.: A day in third grade: a large-scale study of classroom quality and teacher
and student behavior. Elem. School J. 105(3), 305–323 (2005)
Norman, K.: Thinking Voices: The Work of the National Oracy Project. Hodder and Stoughton,
London (1992)
Nystrand, M., Gamoran, A.: Instructional discourse, student engagement, and literature achieve-
ment. Res. Teach. Engl. 25(3), 261–290 (1991)
Nystrand, M.: Dialogic instruction: when recitation becomes conversation. In: Opening Dialogue:
Understanding the Dynamics of Language and Learning in the English Classroom, pp. 1–29
(1997)
236 Y. Song et al.
Nystrand, M., Wu, L.L., Gamoran, A., Zeiser, S., Long, D.A.: Questions in time: investigating the
structure and dynamics of unfolding classroom discourse. Discourse Process. 35(2), 135–198
(2003)
O’Connor, C., Michaels, S., Chapin, S.: “Scaling down” to explore the role of talk in learning:
from district intervention to controlled classroom study. In: Socializing Intelligence Through
Academic Talk and Dialogue, pp. 111–126 (2015)
Podschuweit, S., Bernholt, S., Brückmann, M.: Classroom learning and achievement: how the
complexity of classroom interaction impacts students’ learning. Res. Sci. Technol. Educ. 34(2),
142–163 (2016)
Pöysä, S., Vasalampi, K., Muotka, J., Lerkkanen, M.K., Poikkeus, A.M., Nurmi, J.E.: Teacher–
student interaction and lower secondary school students’ situational engagement. Br. J. Educ.
Psychol. 89(2), 374–392 (2018)
Rampton, B.: Interaction ritual and not just artful performance in crossing and stylization. Lang.
Soc. 38(2), 149–176 (2009)
Reznitskaya, A., Gregory, M.: Student thought and classroom language: examining the mecha-
nisms of change in dialogic teaching. Educ. Psychol. 48(2), 114–133 (2013)
Robin, A.: Developing dialogic teaching: genesis, process, trial. Res. Pap. Educ. 33(5), 561–598
(2018)
Rojas-Drummod, S., Pérez, V., Vélez, M., Gómez, L., Mendoza, A.: Talking for reasoning among
Mexican primary school children. Learn. Instr. 13(6), 653–670 (2003)
Rojas-Drummond, S., Littleton, K., Hernández, F., Zúñiga, M.: Dialogical interactions among
peers in collaborative writing contexts. In: Educational Dialogues: Understanding and Promot-
ing Productive Interaction, pp. 128–148 (2010)
Rojas-Drummond, S., Torreblanca, O., Pedraza, H., Vélez, M., Guzmán, K.: ‘Dialogic scaffolding’:
enhancing learning and understanding in collaborative contexts. Learn. Culture Soc. Interact.
2(1), 11–21 (2013)
Ruiz-Primo, M.A.: Informal formative assessment: the role of instructional dialogues in assessing
students’ learning. Stud. Educ. Eval. 37(1), 15–24 (2011)
Saville-Troike, M.: Extending “communicative” concepts in the second language curriculum: a
sociolinguistic perspective. In: Culture as Core: Interdisciplinary Perspectives on Culture in the
Second Language Classroom, pp. 3–17 (2003)
Sinclair, J.M., Coulthard, M.: Towards an Analysis of Discourse: The English Used by Teachers
and Pupils. Oxford University Press, Oxford (1975)
Smith, M.K., Jones, F.H., Gilbert, S.L., Wieman, C.E.: The classroom observation protocol for
undergraduate STEM (COPUS): a new instrument to characterize university STEM classroom
practices. CBE Life Sci. Educ. 12(4), 618–627 (2013)
Song, Y., Chen, X., Hao, T., Liu, Z., Lan, Z.: Exploring two decades of research on classroom
dialogue by using bibliometric analysis. Comput. Educ. 137, 12–31 (2019)
Soter, A.O., Wilkinson, I.A., Murphy, P.K., Rudge, L., Reninger, K., Edwards, M.: What the
discourse tells us: talk and indicators of high-level comprehension. Int. J. Educ. Res. 47(6),
372–391 (2008)
Vygotsky, L.: Interaction between learning and development. In: Readings on the Development
of Children, vol. 23, no. 3, pp. 34–41 (1978)
Wells, G., Arauz, R.M.: Dialogue in the classroom. J. Learn. Sci. 15(3), 379–428 (2006)
Xie, Z.Y., Cao, X.: Construction three-dimensional observation system of teacher-student interac-
tion behavior in classroom and practice. In: International Conference on Education Technology
& Computer. IEEE (2010)
Application of Parallel Corpus to Teaching Style
and Translation
Lu Tian1,2(B)
1 School of Interpreting and Translation Studies, Guangdong University of Foreign Studies,
Guangzhou, China
ivytianlu@gdufs.edu.cn
2 Center for Translation Studies, Guangdong University of Foreign Studies, Guangzhou, China
1 Introduction
Style is considered the fingerprint of a piece of writing, showing its essential linguistic
characteristics and reflecting the preference and choice of the writer in language use. Such
characteristics can be observed on all aspects—vocabulary, syntax, rhetoric, structure,
etc. In translation, it is important for translators to be aware of the style of the source
text (ST) and then properly retain it or justifiably deal with it otherwise in the target text
(TT). In this process, the translator, as the creator of the target text, may inevitably bring
his or her own style into the translated work.
As style can be observed through repetitive linguistic features, corpus together with
its various retrieval and concordance tools provides an effective method for discovering
and analyzing such features, thus facilitating the discussion and study of the style of
the text. This paper is an account of the application of the Chinese-English Parallel
Corpus of Hong Lou Meng to the teaching of translating narrative markers, a key style
of Chinese vernacular fictions, in the module of Style and Translation for postgraduates
of Translation Studies in a Chinese university.
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 237–245, 2020.
https://doi.org/10.1007/978-3-030-38778-5_26
238 L. Tian
“Huashuo” is dealt with in more variety in H’s where it is presented two dozen
times in the TT, roughly accounting for a quarter of the total occurrences in the ST. No
specific rule is found in terms of the translator’s selection of which ones to translate and
which ones not to. In other words, it seems that Hawkes & Minford were quite casual
in picking narrative markers for translation. Despite this defect, their endeavor to show
the variety of language can be readily discovered. Generally, H’s expresses the meaning
of “huashuo” from three points of view—“we”, “you”, and “story”. Table 1 shows the
detail.
240 L. Tian
Table 1 shows that in rendering the narrative marker “huashuo”, H’s adopts expres-
sions in three categories as the subjects of clauses. The starting point of information shifts
from “we”, which refers to the author and the reader or the narrator and the narratee, to
“you”, simply the reader or narratee. Another alternative is to start with reference to the
“story” by employing subjects including “our last chapter”, “our story” and the pronoun
“it”. As far as predicators and the associated processes are concerned, H’s again adopts a
variety of devices, including material, verbal, and mental processes, in flashbacking the
content of the previous chapter despite the fact that “huashuo” is unanimously verbal in
the ST.
It is worth noting that in some chapters H’s capitalized each letter of the very first
word, for instance “WHEN Caltrop saw the cousins…”. However, the translator did not
follow this practice throughout the whole translation, and hardly could any criteria be
figured out as which ones to capitalize and which ones not to. Moreover, in two chapters
the text begins with “TO CONTINUE OUR STORY”, an all-letter-capitalized phrase to
function as the beginning markers.
In comparison, J’s is the most significant in that it, in one way or another, retains the
meaning of “huashuo” in all the chapters where it appears in the ST. Its renditions are
summarized in Table 2.
As is shown in Table 2, J’s mainly takes two categories of subjects in rendering
“huashuo”—“we” referring to the narrator (author) and narratee (reader), and “story” in
diverse forms. As to the predicators, most predicators present the material process. With
a detailed look at the goals of these processes, it is found that they mainly fall into two
kinds—“our story” and “our narrative”. Both explicitly reflect the story-telling origin of
Chinese vernacular fictions and ensure the continuity of narrative. In addition, similar
to H’s, J’s also employs “to-infinitive” to start a chapter. Examples of such expressions
include “But to proceed/continue”, “But to resume/return to our narrative/story” and
Application of Parallel Corpus to Teaching Style and Translation 241
“But to return to our narrative”. In all, expressions like these are employed up to ten
times in J’s in conveying the meaning of “huashuo”, which is much more than that in
H’s despite the fact that J’s only covers the first 56 chapters of the 120-chaptered novel.
In summary, among all the three translations of the beginning marker “huashuo”,
J’s stands out in conveying its narrative function and having all of them translated. Both
H’s and J’s adopt a variety of language devices in expressing “huashuo” in English and
sometimes employ additional explanatory words for narrative, which makes the TTs not
as concise as the ST.
As the counterpart of beginning markers, ending markers are used to mark the ending
of a chapter or an event. At the end of each chapter in Chinese vernacular fictions, there
is usually an intentional suspense imitating the end of an oral performance, which aims
to arouse the curiosity and interest of the audience so that they will come back and pay
for the performance.
“Xiahui fenjie” (next chapter explain) is a typical ending marker, appearing 106
times in HLM. Along with this ending marker, the narrator sometimes intentionally
drops some hints on what will happen next. The way to translate “xiahui fenjie” in the
three translated texts is various, reflecting the translators’ understanding of the textual
and narrative functions of narrative ending markers. In H’s, literal translation is adopted,
242 L. Tian
and the sentence structures vary as the contextual situation changes. While there is almost
no ellipsis in translating this set phrase, H’s seldom presents any narratological nature
of this ending marker.
In Y’s, a variety of linguistic devices including conditional clause, passive voice and
modal verbs indicating strong degree of imperative are adopted. For instance, the ending
marker is translated as “If you want to know what followed, read the next chapter”
and “To know whether she lived or died, you must read the next chapter.” The main
clauses in both sentences are imperative. The use of modal verb “must” in the second
sentence indicates a strong suggestion. The subordinate clause in the first sentence and
the adverbial phrase in the second are respectively conditional and purposeful, which on
the one hand function as connections between chapters in content, on the other hand play
the role of arousing the audience’s curiosity. In this sense, Y’s does well in preserving
the narratological characteristics of Chinese vernacular fictions through the translation
of narrative ending markers.
In J’s rendition of “xiahui fenjie”, appellation and interrogative are adopted, which
highlights the interpersonal function of the narrative expression. For instance, the trans-
lation of “qieting xiahui fenjie” in J’s is “But, reader, listen to the explanation contained
in the next chapter.” The use of “reader” and “listen to” indicates both the interpersonal
relationship between the writer and the reader as well as that between the story-teller and
the listener in a pseudo-story-telling scene, reflecting the legacy of storying-telling tra-
dition in Chinese full-length vernacular fictions. Moreover, interrogative is widely used
in J’s, which expresses a strong degree of appeal to the audience. For instance, “But
reader, do you want to know the sequel?” In this sense, J’s successfully reproduces the
pseudo-storytelling scene which traditional Chinese vernacular fictions originate from.
Just as their functions in the ST, the ending markers in J’s play the role of summarizing
the previous story and arousing the curiosity of the audience.
untranslated and H’s two. J’s re-presents all the twenty-two markers in the first fifty-six
chapters.
Figure 2 shows the concordance of “yiyu weiliao” and its translations sorted by
ParaConc, which greatly enhances the accuracy and efficiency of data retrieval. For
easier observation and comparative analysis, the sorted data was exported and opened
in Excel with clearer display of the parallel data. Figure 3 is a snapshot of the parallel
texts of “yiyu weiliao” and its translations shown in Excel.
In both Y’s and H’s, the adverbial phrase “just then” is adopted ten and four times
respectively in rendering “yiyu weiliao” to indicate the time of event, making the TTs
look brief. In addition, H’s uses the prepositional phrase “(just) at that (very) moment”
fourteen times, much more than the only one adoption in Y’s. Both expressions have sim-
ilar functions and pragmatic effects. It is interesting to find that “at once” is coincidentally
244 L. Tian
Fig. 3. Parallel texts of “yiyu weiliao” and its translations shown in Excel
adopted in both translations in handling a same text, which shows the influence of con-
text on translators’ choice. Other mostly adopted expressions include adverbial clauses
of time with words such as “as”, “while”, and “before” as the conjunctions. Generally
speaking, H’s and Y’s show great similarity in the rendering of “yiyu weiliao”.
J’s way of handling “yiyu weiliao” is unitary. They are all put into adverbial clauses.
Among its twenty-two occurrences, sixteen are put in inverted order. Adverbs such as
“scarcely”, “hardly” and “barely” are alternatively adopted with frequencies of nine, nine
and four respectively. Although renditions as such make the texts appear a bit redundant,
J’s best presents the narrative function of turning markers in bringing sudden intrusions
into the scene in focus.
3.4 Summary
In translating Hong Lou Meng, though the omission of narrative markers may not nec-
essarily make much difference to the flow of the story, it will run the risk of losing a
key style of this Chinese classical vernacular fiction in terms of narration. After detailed
comparative analysis of the three English versions, it is discovered that all the three
English versions have their gains and losses in the translation of the narrative markers
under discussion. With the aim to help foreign readers learn the language, J’s excels
in the proportion of narrative markers retained in the translated text whereas the other
two versions only keep less than a half. As far as the specific translation methods are
concerned, Joly is found to prefer the use of clauses, so the TT tends to be long and
redundant. This is in accord with the fact that J’s overall length is longer than the other
two. In comparison, the language of Y’s is the most concise. As it focuses more on the
function of expressions, Y’s is not confined to the literal meaning of narrative markers.
Application of Parallel Corpus to Teaching Style and Translation 245
H’s pays attention to the diversity of language use, so there is usually more than one
way to translate a same narrative marker.
4 Conclusion
Acknowledgements. The work was substantially supported by The National Social Science Fund
of China (Project No. 19BYY125).
References
1. Liu, Z.Q., Tian, L., Liu, C.P.: The compilation of Hong Lou Meng Chinese-English parallel
corpus ( ). Contemp. Linguist. 10(4), 329–339 (2008)
2. Zhao, H.H.: The Uneasy Narrator: Chinese Fiction from the Traditional to the Modern. Oxford
University Press, Oxford (1995)
3. Leech, G., Short, M.: Style in Fiction: A Linguistic Introduction to English Fictional Prose,
2nd edn. Pearson Education Limited, Harlow (2007)
4. Liu, Z.Q., Tian, L.: Narrative markers in Hong Lou Meng and their translations—a corpus-based
study ( ). Foreign Lang. Res.
1, 106–110 (2009)
5. Chen, P.Y.: The Conversion of Narrative Patterns of Chinese Novels (
). Shanghai People’s Publisher, Shanghai (1988)
6. Yu, X.H.: Research on the Classical Vernacular Fictions ( ). Anhui People
Press, Hefei (2005)
7. Plaks, A.H.: Chinese Narrative. Peking University Press, Beijing (1995)
The Design and Application of an Web-Based
Online Examination System
Jilu Jiang1 , Baoxian Wu2 , Liang Chang1 , Kui Liu3(B) , and Tianyong Hao2
1 Audit Governance and Risk Control Research Center, School of Accounting,
Guangdong University of Foreign Studies, Guangzhou, China
jiangjl@gdufs.edu.cn, changlianggdufs@163.com
2 Guangzhou Key Laboratory of Big Data and Intelligent Education, School of Computer
Science, South China Normal University, Guangzhou, China
876354386@qq.com, haoty@m.scnu.edu.cn
3 Network and Modern Educational Technology Center,
Guangzhou University, Guangzhou, China
liukui@gzhu.edu.cn
Abstract. Online examination has been used more and more widely in the educa-
tion and other fields due to its advantages of efficiency, convenience, and fairness.
This paper proposes a new web-based online examination system, which utilizes
PHP, Ajax and other technologies to implement online examination functions,
including testing question collection, user management, online testing, real-time
score calculation, answer checking, and result analysis. Particularly, an automated
test paper generation module was designed and integrated. The system has been
applied to a course involving more than 1000 students per semester at Guangzhou
University of Foreign Studies. It has been proved to save efforts of teachers and
students, demonstrating its effectiveness in assisting teaching and learning.
1 Introduction
results, rolled out textbook randomly and separated teaching from testing. All of exist-
ing researches reflect the advantages and the tendency of utilizing online examination
systems in the internet and technology era.
Through the investigation of existing online examination systems in Chinese main-
land, some situations have been discovered as follows: The online examination systems
are mainly utilizing general computer-based techniques. Most of test papers are created
by utilizing a random sampling strategy. At the end of examination, scores are recorded
without any further deep analysis. Therefore, based on the investigations, this research
tries to design an online examination system based on a new strategy for automated test
paper generation. This system is characterized by a list of features including: (1) The
examination system is applied to the evaluation of accounting course learning. (2) The
system serves not only for examinations but also for periodically exercises. (3) Ques-
tion banks are created based on a knowledge graph. (4) A new automated test paper
generation strategy is proposed and implemented.
2 System Architecture
System requirements are firstly analyzed to ensure reliability and safety. The system
mainly uses a Model-View-Controller (MVC) framework to separate data processing,
data representation, program input and output controls. In the framework, controller is
responsible for forwarding requests and processing requests, and view is for graphical
interface design, while model implements functions and algorithms, data management
and database design. The framework not only makes system program structure clear
and flexible but also reduce the degree of coupling among functions. Moreover, the
framework enhances dynamic programming and simplifies subsequent modifications
and expansions of the system.
Based on the analysis, the system is designed with a list of modules. Each of the
modules corresponds to a list of functions. The overall architecture of the online exami-
nation system design is shown as Fig. 1. In addition, this system is developed and tested
in WAMP general application environment, which is a set of open source softwares that
are commonly used to build websites or services. WAMP contains an integration of
Apache, Mysql, MariaDB, Perl, PHP, and Python.
In the system architecture, there are three user centers associating with a list of modules. A
student user center includes personal information maintenance, examination participant,
and score view. A teacher user center includes personal information maintenance, student
management, class management, and so on. An administrator user center contains all
the modules of teacher user center with additional modules such as teacher management
and class management. In general, the major modules of the system consists of student
management, class management, question bank management, examination management,
automated test paper generation, score management, teacher management, etc. Some of
the essential modules are described as follows, and the automated test paper generation
module is specifically introduced in Subsect. 2.2.
The Design and Application of an Web-Based Online Examination System 249
Landing page
Users
Users
Automated test
paper generation
(1) Student management module. Administrators or teachers can insert, edit, delete
and search students in the system using the module. Each student has basic information
including student number, user name, associated class, courses, etc. Since it is time-
consuming and error-prone to add students to the system one by one, the system provides
a convenient way by reading all students’ information from a batch of file. After logging
in the system with a user name and password, a student can view personal user center to
maintain his/her information including password reset, personal icon modification, etc.
(2) Question bank management module. The question bank contains four types of
objective questions including blank filling, single-choice, multiple-choice, and true-
false. Permitted users can search, add, modify and delete questions. Each question con-
tains a question title, multiple candidate answer options, correct answer, as well as a
difficulty level and its coverage in corresponding textbook. Considering teachers may
not have enough database knowledge, the system also provides a friendly user interface
to operate the module intuitively and conveniently.
(3) Examination management module. Teachers or administrators can initiate, modify,
suspend, and terminate an examination in the module. A teacher can indicate the name
of the examination, duration, coverage of textbook, students, score sheet, etc. to create
a new examination. He/she can then design a detailed test paper manually or using the
automated test paper generation function. After that, the teacher can suspend and modify
the examination information. When students log into the examination system, they can
find a list of currently available examinations and choose one to start an online test.
(4) Score management module. When a student completes an online examination, the
system calculates a final score automatically by comparing student’s answers with correct
answers and displays the score to the student. At the same time, the user can choose to
view the correct answers or view the wrong questions for learning improvement. The test
250 J. Jiang et al.
score, associated with students’ information, and examination information are recorded
into the database. Teachers and administrators can view overall scores, sort student by
scores, search the score of a particular student, modify scores, and output scores in a
certain format. By measuring the scores, a teacher can adjust his/her teaching plan and
teaching scheme accordingly to improve teaching performance.
After that, the module organizes the selected questions by following some rules to gen-
erate a test paper automatically. For instance, the total of single-choice questions are 20
thus 10 questions are needed.
Question bank
built
Knowledge graph
creation
Test paper
Automated test
requirement
paper generation
design
Examination
Achievement
analysis
Fig. 2. The process of automated test paper generation in our online system.
3 System Implementation
The usage of system functions needs user permission and authorization. When a user
tries to login the system, account name, password, user role, and verification code are
required. When he/she enters the information in the login page, the system sends an
Ajax request to the server to verify the account, password, verification code and other
information. If the error happened, an error message is displayed on the login page. If
user information is correct, the login is successful and the corresponding user interface
is loaded and displayed.
In order to prevent malicious attacks, the system uses the verification code technol-
ogy in the user login interface. Nowadays, there are many kinds of verification codes
on the Internet, such as characters, voices, and question and answer. Users can visually
recognize verification code information and input the form to submit the website for
verification. In our system, a more straightforward method is used to generate a 4-digit
verification code. Although the system login test and other problems become trouble-
some after the introduction of the verification code, this feature is necessary for system
security. Besides, users can tick “remember me” function when logging in so that user
password information is stored in the cookie for 24 h and no repeated password input is
needed for the convenience of user login.
252 J. Jiang et al.
Teachers and system administrators can enter the system backend to manage data
after user authorization on the login interface. Due to different permissions of teach-
ers and administrators, all data management modules are separated. In the backend,
teachers can perform student management, class management, question bank manage-
ment, examination management, score management, and maintain personal information.
In addition to the modules, system administrators can add, delete, and modify teacher
modules.
(1) Student management module. System administrators and teachers have the right to
use the student management module. Figure 3 shows a screenshot of the user interface
of the student management module.
System administrators or teachers can add or delete student information through clicking
the add button on the student management page. When they input a student’s name,
student number, and password, select a class, and click the Add button, the information
of a student can be added through Javascript code verification. For student deletion, in the
operation of the last column of student records on the student management page, there
is a delete icon corresponding to an Ajax request sending to remove the corresponding
record. At the same time, the system supports the function of batch deletion, that is,
deleting multiple pieces of data at the same time by clicking “checkbox” before the
relative records.
(2) Class management module. System administrators or teachers have permission to
enter the module for the management of class information. They can add classes, delete
classes, modify class information, or search for classes with the input of a class name.
The detailed operations of the management actions are similar to that of the student
management module.
(3) Question bank management module. Question bank management model consists of
single-choice questions, multiple-choice questions, true-false questions, and blank filling
questions. Taking single-choice question management as an example, a screenshot of the
The Design and Application of an Web-Based Online Examination System 253
user interface is shown as Fig. 4. Teachers or system administrators could add or modify
single-choice questions, delete or bulk delete single-choice questions and search for
single-choice questions based on single-choice questions or chapters. Multiple-choice
questions management and true-false questions management are similar to that of single-
choice management.
Fig. 4. A screenshot of the single-choice question management in the question bank management
module
(4) Examination management module. System administrators and teachers have full
access to the examination management module. As conventional operations, they can
add examination information, delete an examination, modify examinations, or search
for examinations based on their names. A screenshot of the examination management
module is shown as Fig. 5.
(5) Score management module. System administrators and teachers have full access
to the score management module. Similarly, they can search or modify students’ scores
on the score management user interface. Meanwhile, they can view some statistical
analysis of students’ scores including the average scores for every question, and the
changes of student’s performance on every knowledge point compared with that in the
past examination, etc.
(6) Teacher management module. System administrators have permissions to enter the
teacher management module and they can add, delete, modify teacher information, or
search for teachers based on the basic information of teachers.
(7) Examination module. After user information authorization, students can enter the
user-side of the module to take examinations, maintain personal information, and check
individual examination scores. Figure 6 shows the user interface of an ongoing online
examination. The page displays the information of examination name, time requirement,
test questions in single-choice, multiple-choice, blank filling and true-false types, etc.
When a student completes the test paper, he/she can click the submit button and the
system automatically calculates the score of the test paper automatically in real time.
254 J. Jiang et al.
The page also has an automated time counting-down function with JavaScript. The test
paper will be automatically submitted when the examination time is used up.
4 Conclusion
As the wary of learning change, more and more people acquire the knowledge they
need through internet. Examination systems have gradually transformed from means
based on traditional printed-paper to the online one. This paper introduces a new online
examination system containing an automated test paper generation method that has been
The Design and Application of an Web-Based Online Examination System 255
applied to a university with more than 1000 students per semester for two years. The
system has implemented most of the major functions and modules for the completion of
an online examination systems. However, it still has some limitations such as the lack
of automatic scoring for objective questions, for this is a highly challengeable and open
problem. More efforts will be made to improve its capability to adapt to different needs
of online examinations in the future.
References
1. Zhu, J.W.: Implementation of an online examination management system based on .NET
framework. Master’s thesis, Chengdu University of Electronic Science and Technology, 7–11
(2009)
2. Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R. (eds.): SETE 2018. LNCS, vol. 11284.
Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03580-8. ISSN 0302-9743, ISBN
978-3-030-03579-2
3. Chen, X., Hao, J., Chen, J., Hua, S., Hao, T.: A bibliometric analysis of the research status of
the technology enhanced language learning. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau,
R. (eds.) SETE 2018. LNCS, vol. 11284, pp. 169–179. Springer, Cham (2018). https://doi.
org/10.1007/978-3-030-03580-8_18
4. Chen, X., Wang, S., Tang, Y., Hao, T.: A bibliometric analysis of event detection in social
media. Online Inf. Rev. 43(1), 29–52 (2019)
5. Song, Y., Chen, X.L., Hao, T.Y., Liu, Z.N., Lan, Z.X.: Exploring two decades of research on
classroom dialogue using bibliometric analysis. Comput. Educ. 137(2019), 12–31 (2019)
6. Wang, X., Hao, T.: Designing interactive exercises for corpus-based English learning with
Hot Potatoes software. In: Huang, T.-C., Lau, R., Huang, Y.-M., Spaniol, M., Yuen, C.-H.
(eds.) SETE 2017. LNCS, vol. 10676, pp. 485–494. Springer, Cham (2017). https://doi.org/
10.1007/978-3-319-71084-6_57
7. Dun, Y.J., Wang, N., Wang, M., Hao, T.Y.: Revealing learner interests through topic mining
from question-answering data. Int. J. Distance Educ. Technol. (IJDET) 15(2), 18–32 (2017)
8. Ren, D.Y.: Design and implementation of course examination system based on .Net. Shandong
University (2018)
9. Jiang, L.Z.: Development of Java programming online examination system. Comput. Knowl.
Technol. 2019(07), 142–144 (2019)
10. Stergiopoulos, C., Tsiakas, P., Triantis, D., et al.: Evaluating electronic examination meth-
ods applied to students of electronics. Effectiveness and comparison to the paper-and-pencil
method. In: IEEE International Conference on Sensor Networks, Ubiquitous, and Trustworthy
Computing (SUTC 2006), vol. 2 (2006)
11. Pang, G.M., Yuan, J.M.: WEB-based online examination system. Friends Sci. 10(29), 121–122
(2009)
12. Ayo, C.K., Akinyemi, I.O., Adebiyi, A.A., Ekong, U.O.: The prospects of e-examination
implementation in Nigeria. Turk. Online J. Distance Educ. - TOJDE 8(4), 125–135 (2007).
ISSN 1302-6488, Article No. 10
13. Yuan, Z.M., Zhang, L., Zhan, G.H.: A novel web-based online examination system for
computer science education. In: 33rd ASEE/IEEE Frontiers in Education Conference,
S3F-7–S3F-10 (2003)
256 J. Jiang et al.
14. Zhang, L., Zhuang, Y.T., Yuan, Z.M., Zhan, G.H.: A web-based examination and evalua-
tion system for computer education. In: Sixth IEEE International Conference on Advanced
Learning Technologies (ICALT 2006), pp. 120–124. IEEE (2006)
15. Abubakar, A., Adebayo, F.: Using computer based test method for the conduct of examination
in Nigeria: prospects, challenges and strategies. Mediterr. J. Soc. Sci. 5(2), 47 (2014)
16. Ko, C.C., Cheng, C.D.: Secure Internet examination system based on video monitoring.
Internet Res. 14(1), 48–61 (2004)
17. Singh, S.K., Tiwari, A.K.: Design and implementation of secure computer based examination
system based on B/S structure. Int. J. Appl. Eng. Res. 11(1), 312–318 (2016)
18. Cen, G., et al.: A implementation of an automatic examination paper generation system. Math.
Comput. Model. 51(11–12), 1339–1342 (2010)
A Text Mining Application in Operation
Management Course Teaching
caoxy@cnis.ac.cn
Abstract. Sharing bicycle, as one of the hottest and newest industries in recent
years, has drawn much dramatic attention from society. In the operation manage-
ment course in this paper, students are expected to analyze the interview texts and
investigate the problems occurred in sharing bicycle with corresponding sugges-
tions. A list of interview texts from sharing bicycle users in Guangzhou Higher
Education Mega Center are collected and analyzed. TF-IDF, as a widely used text
mining method, is applied to extract frequently used key words for a qualitative
analysis. Finally, ten key problems are identified and summarized, which provide
government suggestions about supervision, such as user-centered management,
user experience improvement, user interest protection, and deposit management.
1 Introduction
The sharing bicycle is an emerging transport that provides users with another environ-
mentally friendly and convenient travel option to make up for the slow and crowded
traditional public transport [1]. According to the incomplete statistics from the Ministry
of Transport, there are more than 16 million sharing bicycles on the market from 2010 to
2018 [2]. The number of users of sharing bicycles in China was 28 million in 2016, but
increased to 235 million in 2017, with a growth rate of 632.1% [3]. The explosive growth
not only drives China’s sharing economy, but also expands the global sharing bicycle
market. According to Cheetah Data, the weekly active penetration rate of the global
sharing bicycle industry increased by 1258% in 2017. From 2016 to 2018, Alibaba, Ant
Financial and DiDi have conducted nearly 10 rounds of financing (nearly $1.5 billion)
to OFO, which is the largest sharing bicycle company in China [4].
However, the massive capital inflows and market competition created fake booms
and bubbles. As more and more cities strengthened the regulation of sharing bicycle
management, its market demand drops abruptly. In 2018, the Beijing Municipal Com-
mission of Transport issued an order banning the sharing bicycles from being on market.
This act was later followed by the cities of Shanghai, Guangzhou and Shenzhen. As the
market shrank in 2018, the problem of excess capacity among bike suppliers is gradually
exposed. In the first half of 2018, a large number of sharing bicycle enterprises went
bankrupt. The operating income of OFO Company decreased by 82.3% [5].
In the context of a shrinking market, the remaining sharing bicycle companies are
exploring the appropriate direction for future development. Therefore, the OFO, as a
representative company, is frequently taken as an example in the course teaching of
operation management to help students understand the existed problems in the opera-
tion management of sharing bicycle industry, which plays a crucial role in enterprise
transformation learning.
Instead of focusing on the identification of key factors in sharing bicycle operation,
this paper uses real-life interviews as teaching and learning activities to contextualize the
research topic. Exiting research mainly explored the quantitative relationships among
critical factors in OFO operation decision making [6]. For example, Chen and Li (2018)
discussed factors influencing customers’ willingness to use sharing bicycles from the
perspective of rational behaviour theory and a technical acceptance framework model
[7]. Fan and Cao (2018) analyzed the interactive relationship among sharing bicycle users
[8]. However, qualitative research on sharing bicycle remains extremely rare. Therefore,
the operation management course teaching is used in this research and students are
required to use qualitative methods to analyze interview texts from OFO users in the
local communities. Through the utilization of a text mining method, students identified a
list of existing problems, which are further summarized and explained in current sharing
bicycle industry.
2 Background
Qualitative analysis clarifies the text structure and the subject relationships inside the
texts through reading, comprehension, interpretation and summary. It relies much on
the researchers’ subjective understanding and processing [9]. Content analysis method,
however, is an analytical method that describes content texts by objective and quantitative
data [10]. It improves the reliability of research by quantitative statistics of text elements
(symbols, punctuation marks, characters and sentences) and the occurrence frequencies
of these elements. Although it brings in the objective indicators, the criteria of evaluation
and classification of contents still involve a great deal of subjective judgments, which can
produce deviations to research results. Content analysis, which is limited to the statistics
of elements, is difficult to explore the deep-seated and hidden relationship among texts
and the text sampling, while computer-aided content analysis bridges the gap.
Computer-aided content analysis relies on the processing and conversion of texts
with the assistance of programming to transform unstructured text data into structured
ones for research and management [11]. It eliminates deviation in the subjective text
comprehension through machine learning and data mining in text similarity, by which
the criteria on text classification and evaluation can be universal. With the development
of the algorithms, the classification accuracy of texts can be improved. In this sense,
computer-aided content analysis method plays a vital role in qualitative research.
A Text Mining Application in Operation Management Course Teaching 259
3 Data
The course reading materials of the OFO operation management are released to stu-
dents, which contain eight randomly selected interviews from sharing bicycle users in
Guangzhou Higher Education Mega Center. The interviews are conducted in Chinese and
the total materials are over thirty pages, with 21984 Chinese characters. The interview
texts are expected to analyze and answer the following questions:
TF-IDF requires a necessary data pre-processing procedure. For English texts, the
following principles are applied: (1) uniform case; (2) remove punctuations; (3) remove
digital symbols; (4) remove short words shorter than the specified length; (5) remove
stop words; (6) stemming, which means the adjective form and the past tense of the
word are transformed to the root of the word. After above steps, the text document is
constituted with a number of unique terms.
Due to the large differences between Chinese and English grammatical structures,
this paper adopts principles of partial data pre-processing before the calculation of TF-
IDF: (1) remove punctuations; (2) remove digital symbols; (3) remove expletives; (4)
remove single words and retain terms that contain more than two words and less than
four words; (5) remove stop words and repeated words in interview materials.
The eight interview documents were analyzed by a software NVivo through manual
encoding. After decomposing and encoding, the following ten coding categories are
determined, as shown in Table 1, including bicycle delivery, bicycle quality, the way of
charge, misuse, privacy protection, changes in life, usage frequency, urban governance,
storage place and deposit management. In the table, reference number refers to the times
of number that this code is occurred among the eight interview texts. The weighted
coverage rate means the coverage rate of a node in NVivo in a single text multiplied by
the coverage rate of the documents containing this node in the total number of documents.
The weighted coverage rate reflects the overall coverage of a node in ten documents and
is calculated by applying Eq. (4).
|{a : Ci ∈ dc }| Ri
Rwc = (4)
|D|
Rw,c is the weighted coverage rate of a particular node, and |{a : Ci ∈ dc }| is the
number of documents that contain the node. Ri the sum of coverage rate of the node
in all documents. The denominator |D|. is the number of documents in D.
The identified words and their frequencies analyzed by Nvivo are shown in Table 2.
The shortest term length in Chinese is two. “Change” is as the most frequently occurred
A Text Mining Application in Operation Management Course Teaching 261
word that appears 350 times, followed by the word “Government. All of the respondents
expressed a strong desire for government intervention. Combined with the third and fifth
ranked words “Deposit” and “Pricing”, it reflects that the users want the government to
supervise pricing mechanism and deposit management.
However, the results generated by Nvivo are not concentrated enough. There is
dramatic diversity among the 14 ranked words with higher frequencies generated by
Nvivo. In addition, Nvivo cannot filter out stop words from the texts, thus it contain
deviations caused by the preference of each interviewer. Moreover, some long compound
words in Chinese are usually incorrectly identified as multiple short words by Nvivo,
and therefore the results are affected because of the false recognition.
Table 2. The identified words and their frequencies in the interview texts
The TF-IDF value indicates the importance of a word in a single document. Therefore,
the larger the value is, the more important the word is. However, a lesser value of TF-IDF
does not necessarily mean the word is not important. Most of TF-IDF applications are
conducted positively, by which the targeted text analysis is conducted on the result of
maximum value. However, through the reverse application of TF-IDF, we can not only
find out the key theme of eight interview texts, but also test the reliability of the key
points.
A Text Mining Application in Operation Management Course Teaching 263
Four interview texts were randomly taken for TF-IDF computation. Ten words with
the highest frequencies in each interview document were selected. In Table 3, those
terms with negative TF-IDF value were marked in bold type. For the Equation of TF-
IFD calculation, the result can be negative when IDF value is negative. The number of
documents |D| is a constant, which equals to 8. If TF-IDF is negative, it means the tagged
terms are mentioned in all eight interview texts.
From the result, it is found that government intervention and deposit management
are the most vital words to which the students need to pay more attention, in addition to
the theme of the interview - “sharing” and “bicycle”.
Table 3. (continued)
5 Discussion
From the results, it is found that users want government to solve the destruction of market
rules and the disorder of market pricing mechanism caused by vicious competition
among sharing bicycle companies. There are two main factors that lead to vicious price
competition in sharing bicycle industry: (1) some companies dominate the market in
transaction; (2) these advantageous companies lower regulatory standards in market
competition [12]. Sharing bicycle services failed to differentiate among companies. The
market segmentation and market positioning are also quite similar with each other, thus
the strategy of corporate mainly focuses on how many bicycles can occupy the market
quickly. This strategy uses capital erosion of other competitors’ market share to achieve
higher returns and more funding in the future, especially when it survives as a monopoly
in the market [13]. However, as the funds are mainly used for marketing, the daily
maintenance are largely cut. As a result, obvious regression occurs in the management
of the sharing bicycle industry. This is the reason why the problems mentioned in the
interview occurred, in which the maintenance of damaged bicycles cannot be kept up,
and the bicycles are parked anywhere without being recycled.
As the society steps into the era of big data, users pay more attention to their personal
privacy information, which can bring unprecedented value to enterprises. There are four
A Text Mining Application in Operation Management Course Teaching 265
ways to disclose users’ personal privacy information in company’s daily operation: data
collection, secondary use of data, misuse of data, and unauthorized use of data [14]. Due
to the demand of rent price calculation, sharing bicycle enterprises have an advantage
in collecting and tracking customers’ data. If customers want to use the sharing bicycle
service, they have to exchange their private information in transaction. Consequently, the
results reflect that users hope the government to strengthen the supervision on enterprises’
collection and usage of personal privacy information.
The results point out the necessity of government’s supervision on deposit. The lack of
regulation and guidance on deposit in sharing bicycles industry are caused by insufficient
supervision of relevant regulatory authorities [15, 16]. The sharing bicycle industry has
been developed rapidly in China since 2017. The expansionary speed with capital support
is far beyond the prediction of government departments and financial regulators. Due to
the immature financial technology, when it comes to the deposit transactions with fast
turnover and large amount of information, the relevant departments do not have effective
means to supervise the transaction data.
Although it is reasonable for enterprises to charge users’ deposit as a pledge, many
users are still concerned about the safety of the deposit. That the users pay the deposit as
a pledge does not mean the users agree with how-to-use the deposit by enterprises. Users
are concerned that the usage of deposit will change the ownership, since the enterprises
make profits by investing the deposits. However, the risk of loss will be transferred to
the users if the investment fails. If the deposit does not have the function of currency
circulation, it should have not been flowed into the market [17, 18]. As an asset, the
flow of deposit can create more value. In this regard, as long as the use of deposit is
reasonable and legal, and an agreement has reached between enterprises and users on the
ownership and the right to use, the deposit flowing into the market will have a positive
effect [19].
6 Conclusion
As one of the hottest and newest industries in recent years, sharing bicycle is often used
in the course teaching of operation management. In this paper, a text mining analysis on
interview text data from users of OFO sharing bicycle in Guangzhou Higher Education
Mega Center was conducted. The highly frequently used words were identified by using a
TF-IDF method to investigate the existing problems occurred in sharing bicycle industry.
A number of key problems including government supervision and deposit management
are summarized and analyzed to provide policy suggestions to the government.
Acknowledgements. This paper is supported by the science and technology plan of Guangzhou
(No. 201804010296) and Natural Science Foundation of Guangdong Province, China. (No.
2018A030310051).
266 Y. Qu et al.
References
1. Singla, A., Santoni, M., Bartók, G.: Incentivizing users for balancing bike sharing systems.
In: AAAI 2015 Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence,
pp. 723–729 (2015)
2. Su, J.: Sharing economy: drivers, problems and prospects. J. Xinjiang Normal Univ. 39(02),
126–131 (2018). (Philosophy and social sciences edition)
3. China sharing economy development report (2019). http://www.sic.gov.cn/News/557/9904.
htm
4. Ai Media: Special research on 2018 China’s Sharing bicycle development status report (2018).
http://www.iimedia.cn/63243.html
5. Li, K.: Study on the development countermeasures of urban shared bicycle from the
perspective of sharing economy. City 3, 66–69 (2017)
6. Li, M.: Marketing research on Shared Bicycle. Financial Industry (academic edition) 5, 121–
123 (2017)
7. Chen, C., Li, X.: A study on the influencing factors of citizens’ willingness to use Shared
bicycles. J. Manag. 15(11), 1601 (2018)
8. Fan, Y., Cao, Y.: Sharing of legal responsibilities among users under sharing mode-a case
study of sharing bicycles. Tianjin Law 3, 1–6 (2018)
9. Qiu, J., Zou, F.: Research on content analysis. Chin. Libr. J. 2, 12–17 (2004)
10. Noble, H., Smith, J.: Qualitative data analysis: a practical example. Evid. Nurs. 17(1), 2 (2014)
11. Marshall, H.: Horses for courses: facilitating postgraduate research students’ choice of Com-
puter Assisted Qualitative Data Analysis System (CAQDAS). Contemp. Nurse 13(1), 29–37
(2002)
12. Tan, Y.: Exploration and prevention of “bottom line competition” in shared bicycle. Price
Theory Pract. 3, 36–40 (2017)
13. Zhang, Y., Liu, X., Zhang, H.: Urban public bicycle system in China: current situation,
problems and countermeasures. China Dev. 13(5), 74–79 (2017)
14. He, M., Liang, X.: Can the system and mechanism of sharing platform promote consumers’
continuous willingness to share? – influence mechanism of institutional trust on shared
platforms. Financ. Econ. Rev. 236(8), 75–84 (2008)
15. Deng, D., Li, Z.: Study on the nature and supervision of shared bicycle deposit management.
J. Southwest Jiaotong Univ. 18(04), 94–100 (2017). (social science edition)
16. Guo, P.: Bike-sharing: collaborative governance in Internet technology and public services.
J. Pub. Adm. 14(3), 1–10 (2017)
17. Liu, Y.: Analysis on the development of Shared Bicycle. Financ. Times 8, 251 (2017)
18. Wu, M., Cheng, N., Li, L.: Public welfare and business of shared bicycle from the perspective
of rent and deposit management. Price Theory Pract. 5, 136–138 (2017)
19. Song, J.: An analysis on regulatory issues of deposit management for shared bicycles. Gansu
Finance 4, 27–30 (2017)
Leveraging Neural Network for Online Learning
Performance Prediction and Learning
Suggestion
Yingshan Shen1 , Weiwei Liu1 , Qiumei Wu1 , Ruiyang Chen1 , and Kui Liu2(B)
1 School of Computer Science, South China Normal University, Guangzhou, China
shenys@m.scnu.edu.cn, 674960668@qq.com, 1014894992@qq.com,
243189663@qq.com
2 Network and Modern Educational Technology Center, Guangzhou University, Guangzhou,
China
liukui@gzhu.edu.cn
Abstract. Learning performance analysis is such a research field that draws much
attention from researchers though it has just been emerged in recent years. On the
one hand, analyzing learning behaviors can help learners to choose their learning
methods and allocate their study time in a more appropriate way. On the other hand,
learning analysis can provide valuable feedbacks for teachers and administrators
to improve teaching efficiency and quality. This paper studies and analyzes more
than 640,000 learning data from the MOOC platform edX. A tree-based model
along with an information gain measure is applied to identify the usefulness of
data features. A back-propagation neural network model is further adopted to
train data and achieve a prediction model of learning performance. In addition,
a genetic algorithm calculates learning score conditions and return feedbacks as
suggestions to learners. Experiment results demonstrate the effectiveness of the
utilization of the methods in the predication of online learning performance.
1 Introduction
With the fast development of information technology in education area, dramatically
increasing of educational data has been accumulated. The data has attracted more and
more attentions from both academic researchers and industry engineers as the data is
an essential source to measure the quality of education and analyze learning behaviors.
For example, Song et al. utilized a bibliometric analysis to explore the research on
classroom dialogue in the past two decades [1]. By analyzing a large number of data
collected during the teaching and learning processes, various learning styles of learners
and different situations of learning environments can be discovered [2]. These findings
can further reveal the significance of improving learning efficiency through feedbacks
and predictions of learners’ learning trends [3]. There are a variety of online education
platforms, e.g., Moodle, Blackboard, MOOC, etc., containing a large amount of learning
record data including the information of learning behaviors, emotional states, facial
features, cognitive activities, attention levels and so on. The data sources for learning
analysis are widespread and the data types are extraordinary diverse [4].
Learning analysis technology can be used to evaluate the quality of online courses
and performances. For teachers and administrators, it provides more specified methods
and targeted teaching interventions of students on the basis of data analysis [5, 6]. For
researcher, it can be used as an effective tool to find out the learning situations of indi-
vidual students and the utility status of online learning [7]. For technical developers, it
helps to optimize the user interface design of learning management systems, discover the
relationship between frequency and usage paths of each module in a learning manage-
ment system, and develop better functions of systems according to the needs of software
engineers to carry out appropriate learning analysis technology [8].
Among the learning technology, machine learning algorithms are frequently applied
to learning analysis and data prediction [9]. Recently, new machine learning algorithms
particularly deep learning methods are receiving more and more attentions in domain
applications. For example, Back-Propagation (BP) neural network, as one of artificial
neural networks, is used in many prediction scenarios, such as lottery opening prediction
[10], vehicle speed prediction [11] and its optimization, building insulation material
performance prediction [12], short-term traffic flow prediction [13] to improve chaotic
genetic algorithm, etc. All these applications are oriented to daily life and the BP neural
network has demonstrated suitable performances.
Therefore, this paper tries to apply a BP neural network model for online learn-
ing analysis. In addition, there is a large-scale open online course learning platform
edX, which is created by MIT University and Harvard University in 2012. There are
more than 640,000 learning behavior records, which are initially analyzed by using a
logistic regression algorithm, where the learners’ ability to complete learning tasks is
predicted [14]. Moreover, many literatures report the usage of genetic algorithms for
application optimization. For example, a robust global optimization search algorithm
has been widely used in combinatorial optimization, signal processing, machine learn-
ing and other aspects of predictive controls. Therefore, this paper applies a BP neural
network model for the prediction of learning performance as well as applies a genetic
algorithm for the prediction analysis of learning data for learning behavior suggestion.
2 The Methodology
The approach that analyze education data and make suggestions for online learning is
described in this section. We firstly apply a decision tree algorithm to automatically dis-
covery learning data attributes/features that have different impacts on learning results. A
BP feed forward neural network model is then applied to train a model containing learn-
ing behaviors of learners to find out the correlation between learners’ learning behaviors
and final performance in terms of scores. Based on these correlations and a genetic algo-
rithm, the estimation of learning cost required for the prediction of expected learning
score is implemented. The corresponding goal-oriented learning behavior suggestion is
thus made based on these predicted results. The processes and related methods for the
learning predication and suggestion is descripted below in detail.
Leveraging Neural Network 269
where p(y) is the probability of y occurrence. The smaller p(y) is, the larger of amount
of information contained in y.
Entropy is a measure widely used in information theory that describes the purity
of any sample data set. It is defined as the expected value of information. A data set S
can be divided into m classes, where information entropy is the expected value of the
information contained in a randomly obtained label as Eq. (2).
m
E(S) = − p(yi )lnp(yi ) (2)
i=1
Assuming that E(S aim ) is the entropy of the m class for the target data and S i (i = {0, 1,
2…}) is another entropy of target classification using other classification strategies. The
Gain[i] is the information gain about S aim obtained by classifying data according to data
item S i . When the information gain value is larger, the greater correlation between data
item S i and data item S aim is. Popularly, the information gain Gain is a quantification of
270 Y. Shen et al.
the degree of association between S i data and S aim data. The higher the value of Gain[i],
the closer relationship between S i and S aim is. Assume that n classes can be obtained by
classified data according to S i , the Gain value is calculated using the Eq. (3) as follows:
k
Gain[i] = E(Saim ) − E(Sin ) (3)
n=1
Initialization
Weight learning
Satisfying
Termination
Conditions
End
From the training results of the neural network, final prediction score is calculated
based on seven aspects, such as the number of learning chapters, the number of days of
course visits, the number of course interactions, etc. The trained formula of the neural
network is derived by function named BPnet (N), in which N is the data items of the
seven aspects, and the function returns the result value learning grade. Equation (4) is
used to implement the predication of certain expected feature values.
In the equation, A refers to the known data items, grade refers to the value of target
score, X refers to an expected feature, N is a set of factors that affect the final score
value in the trained neural network. N is composed of A and X. From the trained neural
network BPnet, the genetic algorithm GA is computed. The approximated feature value
achieved with the goal grade is obtained by the recursively genetic algorithm.
272 Y. Shen et al.
When the expected feature X contains multiple items, there are several combinations
of solutions with large deviations in the solution set of X. Furthermore, some of the
combinations deviate from actual needs, which requires a system to repeat siftings the
combination of these solutions in order to get the best suggestion. In this paper, each
data in the training data set is sorted, 1/4 of the first and last data are excluded, and the
average of half data in the middle is obtained as the value of each data (mid). When X
has multiple items, the weight W of each group of solutions is obtained by using the
Eq. (5). n denotes the number of combinations of different solutions, m is the number
of combinations of solutions that are artificially restricted for optimal selection, and k
defines the serial number of data items.
i∈k
Wn = (xn [i]/mid[i]), n = {0, 1, 2..m} (5)
i
The data are from edX1 , which is a MOOC online course learning platform, containing
learning reports from Harvard University and MIT during 2010 to 2013. All the data
contains learning information of participants, including course registered time, the last
login time, the number of course interactions, course visit days, video chips playing
frequency, the end time of each learning chapter, forum posting number, etc. The relevant
personal information of learners, such as course identify, grade, user identify, country,
educational background, date of birth, gender and so on, is also included. A total number
of 640,000 records are used in this research. After filtering out invalid data, such as
incomplete data, 74,000 records are finally used as a data set.
Learning effort refers to the measurable value of behavior that learners put in to
achieve the learning goal of mastering a knowledge point, such as learning time, learn-
ing comprehension ability and other objective factors. It is worth to discuss that how
to minimize the input learning cost and maximize the learning effect. In addition to the
traditional factors affecting the learning cost, some new elements, such as the length of
online learning, the number of questions answered, the timeliness of submitting home-
work, the frequency of online browsing, the frequency of downloading learning resources
and watching teaching videos, the number of posts, etc., are included in the following
experiments.
For the purpose of data mining, the data set is being pre-processed. The preprocessing
consists of the following aspects:
(1) There are 16 disciplines of courses recorded in the edX data set. For the convenience
of data processing, all course identify (ID) belongs to the 16 course disciplines are
renumbered to be 1–16.
1 https://www.edx.org/about-us.
Leveraging Neural Network 273
(2) For the existing missing value cases particularly on the academic qualifications
attribute, different levels of academic qualifications are mapped to a list of sequential
numerics. For example, missing values and incorrect values are mapped to 0, while
the secondary school qualification is set to be 1, secondary school is set to be 2,
and undergraduate is set to be 3, etc.
(3) The null value or missing value of gender attribute is labelled to 0 while male is 1
and female is 2.
(4) Age information is not directly available but it can be calculated from year of
birth. According to the release time of the data, all the year of birth information is
transferred to the actual age of learners. However, there are some false cases, such
as 2011–2013. We therefore set them as a special mark. After that, since 20 years
old is an essential age in this research, all the age lower than 20 is set as 1 and every
increase of 5 years obtains an additional 1 based on the base value.
(5) Others information such as nationality are removed considering that the information
has no practical reference value in this study.
Afterwards, all the data items with 0 or null values are filtered out. A sample raw
data from the edX MOOC platform and the corresponding pre-processed data are shown
in Fig. 2.
Fig. 2. Examples of raw data (top panel) containing learning information from the edX MOOC
platform and the corresponding pre-processed data (bottom panel).
By applying the decision tree method with the information gain measure, as described
in Subsect. 3.1, the edX data can be computed to filter out invaluable attributes. From
the processing result on the edX data set, the attributes of the number of learning chap-
ters, course visits, course interactions, video playback times, course id, forum posts,
274 Y. Shen et al.
education, age and gender have a less impacts on learners’ final performance. Among
them, age and gender have the lowest influences. Therefore, it can be preliminarily con-
cluded that age and gender have little influence on the final learning scores, thus they
are removed from the dataset in the following experiments. The settings and the result
of using the decision tree model is shown as Table 1.
Table 1. The settings of the decision tree model and the result using the model.
In accordance with the BP neural network model, the total of 74,000 data of edX
records is split into three datasets, training data (70%), validation data (15%) and test-
ing data (15%). According to the result of attribute/feature selection, the number of
learning chapters, course visiting days, course interaction times, video playback times,
course id, forum posting numbers and academic qualifications are taken as input condi-
tions. By applying the BP neural network, the learning scores of learners are predicted.
Table 2 demonstrates the predicted learning scores corresponding to each online course
for example learners.
Since the output derivation of the trained BP neural network can be regarded as a
linear problem, the genetic algorithm can obtain satisfactory results when the number
of iterations is larger enough to solve the linear problem. We therefore conduct an
experiment on 1,000 randomly selected samples. The calculated result of error rate after
10,000 iterations closes to 0.
Leveraging Neural Network 275
Table 2. Examples of final learning score predication corresponding to courses for different
learners.
The error histogram is shown as Fig. 3, where Targets represents the actual value,
Outputs denotes the output values calculated using the trained neural network based on
the training data and validation data. Errors is calculated by Targets minus Outputs,
as the horizontal axis in the figure. The vertical line in the middle of the histogram
represents zero error. Accordingly, the distribution of errors on the training, validation,
and test data can be visualized and analyzed.
In addition, to better analyze the errors, error ranges are calculated, as shown in
Table 1. It shows that the result of whole error prediction accounts for the vast majority
within 0.1 (nearly 70%). However, the error is within 0.05 accounting for more than
50% in proportion (Table 3).
From the above experiments, it can be seen that the prediction accuracy using the BP
algorithm is relatively reliable to some extent by viewing the low error values. Thus, it
is feasible to select the features that have great influence using information gain as well
as to predict learning performance in terms of score.
276 Y. Shen et al.
In this processing of research, it is found that so complexity by using decision tree, neural
network and genetic algorithm to build the model which is relatively high. There are
several problems in the prediction and suggestion of learning performance as follows:
(1) There are many factors affecting a student’s test scores, which can’t depend on a
small amount of data recorded to a large extent. On the one hand, in addition to the
recorded learning behavior factors, some sudden situations for individual students
will also affect learners’ performance, then affecting the predicted results.
(2) On the other hand, because of too many factors, many common learning behaviors
or related data that affect learning outcomes cannot be recorded, thus affecting
the comprehensiveness and accuracy of prediction. At present, there are widely
accepted recording standards technology of online learning behavior such as xAPI,
but in fact it is very difficult to use these standards of technology to achieve real
academic performance prediction.
(3) It Lacks of publicly available research and learning datasets. From the point of this
papers, most of the data come from the private data of online learning system or
distance education platform in universities. It is difficult for external researchers
to obtain valuable research data. Although the widespread use of MOOC platform
has created a large number of online learning records, due to privacy protection and
other factors, users’ using data is not fully public. There are only 16 courses on Edx
platform and education data in UCI database are published by Harvard University
and Massachusetts Institute of Technology. However, there are some problems in
these data, such as low data quality and few reference factors.
278 Y. Shen et al.
(4) When the amount of data used in the experiment is small, there may be a large
deviation in the accuracy of the results. This study is based on learning large data
technology for learning and analysis. When the sample size is large enough, it can
significantly improve the prediction accuracy.
4 Conclusions
This paper targets at online learning prediction and learning behavior suggestion for
improving learning performance of learners. A decision tree method with an information
gain measure are applied for data feature selection. After that, a BP neural network
model is adopted for the prediction of learning performance in terms of final learning
scores. Furthermore, with the help of a genetic algorithm, the features that influence
the final scores most and the necessary values for a specific expected learning score
are analyzed and predicted for learning behavior suggestion. Based on 640,000 learning
records of online courses from edX, the experiments are conducted and the results show
that almost 70% of the prediction with errors remain within 0.1, indicating the high
prediction accuracy. Therefore, this work provides a potential way for effective online
learning performance prediction and learning behavior suggestion.
With the rapid development of big data technology, and the opening and collection of
data, it also received wide attention of the mining and analysis of educational data. With
the further development of data collection and openness, the accurate learning prediction
will be more accurate. Therefore, the paper can build learner’s standard learning model
through learning analysis technology, use standardized source data to predict and suggest,
improve the accuracy of prediction results, make more accurate recommendation, and
help learners to conduct personalized learning more effectively, improve the efficiency
of online learning.
References
1. Song, Y., Chen, X.L., Hao, T.Y., Liu, Z.N., Lan, Z.X.: Exploring two decades of research on
classroom dialogue using bibliometric analysis. Comput. Educ. 137(2019), 12–31 (2019)
2. Qu, Y., Yu, Z., Cong, H., Hao, T.: Pedagogical principle based e-learning exploration: a case
of construction mediation training. In: Huang, T.-C., Lau, R., Huang, Y.-M., Spaniol, M.,
Yuen, C.-H. (eds.) SETE 2017. LNCS, vol. 10676, pp. 539–547. Springer, Cham (2017).
https://doi.org/10.1007/978-3-319-71084-6_63
3. Kashiwao, T., Nakayama, K., Ando, S., et al.: A neural network-based local rainfall prediction
system using meteorological data on the Internet: a case study using data from the Japan
Meteorological Agency. Appl. Soft Comput. 56, 317–330 (2017)
4. Lou, W.: The research of high-dimensional data mining technology for big data. Shanghai
University (2013)
5. Chen, E., Heritage, M., Lee, J.: Identifying and monitoring students’ learning needs with
technology. J. Educ. Students Placed Risk 2010(3), 309–332 (2010)
6. Wang, X., Hao, T.: Designing interactive exercises for corpus-based English Learning with
Hot Potatoes software. In: Huang, T.-C., Lau, R., Huang, Y.-M., Spaniol, M., Yuen, C.-H.
(eds.) SETE 2017. LNCS, vol. 10676, pp. 485–494. Springer, Cham (2017). https://doi.org/
10.1007/978-3-319-71084-6_57
Leveraging Neural Network 279
7. Gu, X.Q., Zhang, J.L., Cai, H.Y.: Learning analysis: emerging data technology. J. Distance
Educ. 30(01), 18–25 (2012)
8. Wei, S.P.: Learning analysis technology: mining the value of educational data in the age of
big data. Mod. Educ. Technol. 02, 5–11 (2013)
9. Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R. (eds.): SETE 2018. LNCS, vol. 11284.
Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03580-8. ISSN 0302-9743, ISBN
978-3-030-03579-2
10. Tu, Y.: Application of BP neural network in welfare lottery prediction. Intelligent Computing
Branch of China Operational Research Society. Third China Intelligent Computing Congress.
Intelligent Computing Branch of China Operational Research Society, vol. 2009, no. 4, pp. 16–
19 (2009)
11. Xie, H.: Prediction of driving condition for plug-in hybrid electric vehicles. Chongqing
University, Chongqing, China (2014)
12. Shu, Y., Gu, B.W., Zhang, Y.Q.: Performance prediction of building thermal insulation
materials based on bp neural network algorithm. Build. Energy Conserv. 04, 52–55 (2017)
13. Lun, Z.M.: Short-term traffic flow prediction based on BP neural network optimized by
modified chaotic genetic algorithms. Comput. Program. Skills Maintenance 05, 18–20 (2017)
14. He, C.K., Wu, M.: Analysis and prediction of learning behavior of educational big data based
on edX Platform. Distance Educ. China 06, 54–59 (2016)
15. Li, D.Z., Du, L.Y.: Application of data mining-based student performance prediction.
Heilongjiang Sci. Technol. Inf. 7, 156–157 (2017)
16. Sun, L., Cheng, Y.X.: Research and realization of learning achievement prediction of network
education in the big data era - taking the english examination for undergraduate public courses
as an example. Open Educ. Res. 21(03), 74–80 (2015)
17. He, W.: Examining students’ online interaction in a live video streaming environment using
data mining and text mining. Comput. Hum. Behav. 29(1), 90–102 (2013)
An Empirical Study of Corpus-Based
Translation Teaching in Higher Vocational
Colleges in China
Hefei, China
1 Introduction
Translation studies had not been integrated with corpus linguistics for a long time until
the 1990s. Corpus-based language teaching has become a new teaching mode with the
development of information technology. In English translation teaching, in addition to
the mass storage of teaching corpus, the data emission generated by corpus can also be
archived, counted, analyzed and described quantitatively with the help of a certain tech-
nical support, so as to find out the translation problems that tend to occur during students’
learning, or to find and verify the rules of translation performance characteristics of stu-
dents at a certain stage, or even get the rules of the development of translation ability of a
specific group of students. In addition to the statistics and analysis of the characteristics
of students’ translations, teachers can also combine keyboard recording software such as
Trans log, Input log or Camtasia and other screen capture software to dynamically track
and record students’ translation process, and explore students’ translation behaviors or
the reasons behind their behaviors.
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 280–284, 2020.
https://doi.org/10.1007/978-3-030-38778-5_30
An Empirical Study of Corpus-Based Translation Teaching 281
In the 1990s, there are several trends of translation studies that have shifted from a
prescriptive approach to a descriptive one. These trends inevitably lead to the gradual
development of descriptive translation studies. In this context, corpus-based translation
research has emerged. Mona Baker is a core representative of a group of cutting-edge
scholars who advocated corpus translation research. Baker [1] applied relevant research
results of corpus to translation studies and has made considerable achievements in the
field of translation studies and translation teaching. The application of corpus can not
only drive students’ learning enthusiasm and make classroom teaching subjects diver-
sified, but also reduce the obstacles in translation teaching, encouraging students to
learn translation knowledge better and hence improving their translation ability, and
thereby promote the quality of translation teaching and facilitate the standardization in
translation teaching. It can be seen that the application of corpus in English translation
teaching is not only the need of English curriculum reform, but also the need of stu-
dents to improve their translation competence. In short, the construction of the corpus
of translation teaching can be extended to the scientific research module to study and
promote teaching.
In recent years, the construction of translation teaching corpus has attracted much
attention in this field. One of the corpus-based online translation teaching and learning
platform has been developed by Professor Zhu Chunshen and Dr. Mu Yuanyuan’s research
team. This platform comprises: (1) corpus-construction (including text-annotation, exer-
cises with explanation of answers, and knowledge-based topical boards); (2) the knowl-
edge management system; and (3) electronic program design to interconnect all the afore-
mentioned components for inter-module navigation online [2]. The Platform has been
used in actual teaching of translation and has generated a series of studies [3–5].
The platform relies on the development of bilingual corpus and adopts SAAS multi-
tenant service architecture to ensure the isolation of user data and fill the gap of domestic
and foreign computer-aided translation teaching software. Its corpus is real and vivid,
with complete classification and rich sources. In its platform, the annotated words are
keywords extracted from various text phenomena and translation methods. It is an orig-
inal and strictly defined system used to identify and describe the texts and cultural phe-
nomena that can be used for corpus annotation. There is a total of 200 labeled keywords,
282 W. Zhao and Y. Mu
covering 9 categories currently. Each keyword is strictly defined, and its theoretical sup-
port comes from functional linguistics, textual linguistics, stylistics, translation studies,
etc. The purpose of corpus annotation guided by annotated keywords is to avoid impres-
sionistic comments, and to enhance the teachability of translation methods and skills by
displaying the explanatory power of texts.
Corpora provide resource support for English translation classes in higher vocational
colleges. Corpora contain real language materials used by people in various language
communication situations, provide objective examples of the actual use of language,
intuitively reflect the use of language, and have strong realistic and social characteristics.
In corpus-based English translation teaching in vocational colleges, teachers can avoid
the phenomenon of lacking objective scientific basis by relying on intuition or teaching
experience in traditional translation teaching, which not only fully demonstrates the
characteristics of high-vocational translation teaching, but also provides a lot of real
language materials as support for it.
Compared with the traditional teaching mode of indoctrination, the corpus-based
English teaching in vocational colleges can transform the role of teachers from traditional
knowledge indoctrinator to the guide and organizer of students’ learning of translation.
The “data-driven learning” method proposed by Johns [6] is an advanced computer-aided
teaching method based on corpora. It advocates that the students learn the actual usage
of a word, a sentence pattern or a grammatical phenomenon by observing real linguistic
phenomena and observing, analyzing and summarizing a large number of contexts.
In line with the data-driven learning mode, the online translation teaching and learn-
ing platform, ClinkNotes Online Platform, is adopted by the authors as the empirical
research in the teaching of vocational college courses. With the help of this platform, the
learning process is exploratory, discoverable and independent, which is in line with the
current trend of education. The enthusiasm and initiative of higher vocational students
can be brought into play, and the research thinking and practical ability can be cultivated,
so as to learn translation more effectively.
In this empirical research design, the authors use corpus materials which are closely
related to vocational settings in the teaching for students of different majors. As shown
in Fig. 1, the corpus-based online translation teaching platform includes a corpus with
the classification of various topic themes, which can be mapped with various vocational
fields for students in different majors. For instance, corpus materials related to scenic
spots and tourism industry can be used as vivid teaching samples for students majoring
in tourism management.
The corpus can be marked and saved to suit different teaching purposes. Before the
class, the teachers can release the searched corpus to the teaching platform for students’
preview, or assign the corpus to students in the form of practice text. They can also mark
the corpus of translation skills that need to be explained according to the translation level
of students. During the teaching and learning process, students can click all the markers
on the screen to find the explanations of knowledge points in a certain sentence of the
text (Fig. 2).
An Empirical Study of Corpus-Based Translation Teaching 283
Annotations
Knowledge points
4 Conclusion
As an emerging technology, corpus has a wide application prospect and has made great
achievements in language teaching. The application of corpus in translation teaching is
undoubtedly a great breakthrough in this field. The authors have conducted an empirical
study in the teaching of translation for students of a vocational college, and have gen-
erated the data-driven learning process which focuses on the vocational characteristics
of text material selection and knowledge explanation. This will set an operable teaching
and learning mode for translation teaching and learning in vocational colleges in China.
Acknowledgements. The work was substantially supported by The National Social Science Fund
of China (Project No.19BYY125).
References
1. Baker, M: Corpus linguistics and translation studies: implication and application. In: Baker,
M., et al. (eds.) Text and Technology: In Honor of John Sinclair, no. (1), pp. 233–250 (1993)
2. Mu, Y., Tian, L., Yang, W.: Towards a knowledge management model for online translation
learning. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R. (eds.) SETE 2018. LNCS, vol.
11284, pp. 198–207. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03580-8_21
3. Zhu, C., Mu, Y.: Towards a textual accountability-driven mode of teaching and (self-) learning
for translation and bilingual writing: with special reference to a CityU online teaching platform.
Chin. Translators J. 2, 56–62 (2013)
4. Tian, L., Mu, Y., Yang, W.: Designing a platform-facilitated and corpus-assisted translation
class. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R. (eds.) SETE 2018. LNCS, vol. 11284,
pp. 208–217. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03580-8_22
5. Zhu, C., Wang, H.: A corpus-based, machine-aided mode of translator training: ClinkNotes
and beyond. Interpreter Translator Trainer 5(2), 269–291 (2011)
6. Johns, T.: Data-driven learning: the perpetual challenge. In: Teaching and Learning by Doing
Corpus Analysis, pp. 105–117. Brill Rodopi (2002)
Species Assignment for Gene Normalization
Through Exploring the Structure of Full Length
Article
1 Introduction
Biological experimental results are usually described in published literature. In order to
conduct and interpret their own experiments, researchers need to find the information of
their interest from the research literature. However, with the rapid growth of biomedical
publications, molecular biology has become an information-saturated field. Manually
extracting information from the literature is a time-consuming and labor-intensive pro-
cess. As a result, a major focus of bioinformatics research is to automatically extract
information from published literature, using text mining techniques.
In order to comprehensively annotate gene records and to support queries from
biologists from a variety of backgrounds who may use different names to refer to a gene
of interest, curators of knowledge bases, such as UniProt Knowledgebase (UniProtKB)
[1], need to capture the full range of names and symbols by which a gene is known.
Automatic detection of gene names in the literature and their linkage to gene database
records, also known as gene normalization, is being developed as an alternative to the
time-consuming practice of manual extraction of names. Since gene database records are
species-specific, deciding which species the gene belongs to is an essential and critical
step for gene normalization. Wei et al. [2] conclude that accuracy of species assignment
is critical for the overall performance on the gene normalization task. When the species
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 285–290, 2020.
https://doi.org/10.1007/978-3-030-38778-5_31
286 R. Ding et al.
2 Methodology
In order to investigate how to assign species to genes in full length articles, we studied the
annotations in the full length articles from BioCreative III gene normalization training set
[7], drew observations and developed our algorithm for species assignment accordingly.
2.1 Observations
We drew the following observations, which are helpful for developing the species
assignment algorithms.
(1) Different sections of the article have different roles. For example, the setups of the
experimental study are usually described in the methods section, while the results of
the experimental studies are usually described in results section. Thus, the relevant
species information can be found in the methods section and mentions of the species
do not need to be found in the results section at all. On the other hand, the species
mentioned in the background section may or may not be relevant for normalization
of genes mentioned in the results section.
(2) Many articles describe one or more experimental studies that are focused on genes
from a single species, even when genes from multiple species are mentioned in
the article. Out of the 32 articles we have studied, only 4 articles (PMC 2048754,
2443158, 2579434, and 2631505) conducted their research on genes from more
than a single species. We observed that the genes used in the experiments and their
species are identifiable from the methods section and that such species (if there is
a single one) can be used for species assignment in the results section. However,
if a particular gene’s mentions correspond to different species, we noticed that
authors usually provide that information in immediate context. For example, in PMC
2396500, the experimental study is focused on genes from Arabidopsis thaliana.
When genes from other species are used for comparison purpose, mentions such as
“human DDB2” and “DDB1 in mammalian” were used.
Species Assignment for Gene Normalization 287
(3) For the detection of species from the methods section, we observed that such species
are often mentioned in the titles of the methods subsections, or in the beginning
of the methods section or subsections, where authors introduce how they conduct
their research. Thus, the position (e.g., title of method subsections) and the identifi-
cation of textual patterns involving species in the first 1–2 sentences of the method
subsections can be used to identify the species.
2.2 Algorithm
Based on the three observations, we developed the following species assignment rules
for full length article. The overview of our algorithm is shown in Fig. 1.
For genes that are mentioned in the results section, we first detect the species of the
genes used in the experiments. The detection of species is handled by dictionary match-
ing, using a species dictionary built from NCBI Taxonomy. These species are detected
from the Methods sections and will be called the experimental species. We identify the
experimental species if they appear in: (1) the titles of the methods subsections, and (2)
the first two sentences of the methods subsections and adjacent to phrases such as “sam-
ple”, “derived from”, “carried out”, “harvesting” while extracting the species names. For
example, in PMC 2423616, sentence which describes experimental species is “MEFs
were prepared by harvesting embryonic stage 14.5 mice”.
We described how species being assigned to gene mentions in the results sections
as follow. The most straightforward case is when only one experimental species is
detected. Like the case of major species in our species assignment process for abstracts,
288 R. Ding et al.
this experimental species will be used as the default choice for genes in the result section.
It can be overridden by immediate species context, i.e., in cases where the gene mention
includes (1) species prefix, (2) species in the same noun phrase or (3) has an attached
prepositional phrase with the species name (we will discuss these three rules in detail
below).
If multiple experimental species or no experimental species is detected, then we
hypothesize that the species for a gene mention would be explicitly stated in the results
section. Thus, this situation becomes similar to an abstract which is read before the
methods. Therefore, in this case, each subsection in the result section will be treated like
an abstract and the rules will be applied the same way they were designed for the other
sections. The rules are developed based on our previous work [8], as shown below.
(1) Prefix. If a gene mention has a species prefix, we assign the species based on the
prefix. e.g., ‘AtAurora1’ would be assigned the species ‘Arabidopsis thaliana’.
(2) Same noun phrase. If a gene mention and species are in the same noun phrase,
we assign that species to the gene mention. e.g., ‘Arabidopsis TOC1/PRR1 gene’
would be assigned the species ‘Arabidopsis thaliana.
(3) Attached prepositional phrase. If a species appears in a prepositional phrase that is
attached to the noun phrase containing a gene, we assign that species to the gene
mention.
(4) Species in the same sentence. A gene mention is assigned to a species that occurs
in the same sentence.
(5) Species in the title and in the MeSH terms. If there is only one species in the title
and the MeSH terms, we will assign it to the rest of the gene mentions in the article.
Currently, there is no corpus available for full length article level species assignment. To
evaluate the performance of our method, we use the BioCreative III gene normalization
corpus. The corpus consists of a gold standards test set that includes 50 articles which
are annotated manually. The annotations are in the form of Gene ID and PMCID pairs.
Thus, if there are more than one Gene IDs corresponding to one unique gene name,
there is no information that indicates the species for each of the gene mentions. For this
reason, we cannot evaluate the recall precisely.
Therefore, we run our system and SR4GN on these articles from the BioCreative III
gene normalization gold standards test set and manually analyze the system outputs for
species assignment errors (errors caused by other process, e.g., gene mention recognition,
are not included). Table 1 shows the number of TPs, FPs and FNs of our method and
SR4GN.
We analyzed the errors of our species assignment algorithm, and found that these
errors mainly correspond to cases where no experiment species is detected and no species
information from immediate context can be used. In these cases, some low confidence
rules will be applied, e.g., species in the same sentence. This causes both FPs and FNs.
Other errors involved mentions of homologs, where multiple genes were mentioned
Species Assignment for Gene Normalization 289
together but from different species. Overall, the results indicate that our algorithm yields
good performance in the species assignment process. One possible improvement can be
detecting species name from presence of cell line names. If multiple species are detected
in this manner from all Methods subsections, we can see if specific genes are associated
with them. Techniques used in conjunction with specific genes can also be associated
with species but currently we have not implemented this aspect.
4 Conclusion
We have described a new method of species assignment for full length article. We
proposed a concept, called experimental species, as a default species in the article,
especially in the results section. We also explored the structure of full length article by
treating different sections of full length article differently. Evaluation shows that our
method yields good performance and outperforms SR4GN, a well-known and state-of-
art system that assigns species to gene mentions. We believe our method can be extended
for other text mining tasks that are applied on full length articles. In the future, we plan
to conduct additional evaluation to further illustrate the significance of exploring the
structure of full length article.
References
1. The UniProt Consortium: UniProt: the universal protein knowledgebase. Nucleic Acids
Research, 45(D1), D158–D169 (2017)
2. Wei, C.-H., Kao, H.-Y.: Cross-species gene normalization by species inference. BMC
Bioinform. 12(Suppl 8), S5 (2011)
3. Gerner, M., Nenadic, G., Bergman, C.M.: LINNAEUS: a species name identification system
for biomedical literature. BMC Bioinform. 11, 85 (2010). https://doi.org/10.1186/1471-2105-
11-85
4. Krallinger, M., Leitner, F., Rodriguez-Penagos, C., Valencia, A.: Overview of the protein-
protein interaction annotation extraction task of BioCreative II. Genome Biol. 9(Suppl 2), S4
(2008). https://doi.org/10.1186/gb-2008-9-s2-s4
5. Wei, C.-H., Kao, H.-Y., Lu, Z.: SR4GN: a species recognition software tool for gene
normalization. PLoS ONE 7(6), e38460 (2012)
6. Wei, C.-H., Kao, H.-Y., Lu, Z.: GNormPlus: an integrative approach for tagging genes, gene
families, and protein domains. Biomed. Res. Int. 2015, 918710 (2015)
290 R. Ding et al.
7. Lu, Z., Kao, H.-Y., Wei, C.-H., Huang, M., Liu, J., Kuo, C.-J., Wilbur, W.J.: The gene
normalization task in BioCreative III. BMC Bioinform. 12(Suppl 8), S2 (2011)
8. Ding, R., Arighi, C.N., Lee, J.-Y., Wu, C.H., Vijay-Shanker, K.: pGenN, a gene normalization
tool for plant genes and proteins in scientific literature. PLoS ONE 10(8), e0135305 (2015)
ETLTL (Educational Technology for
Language and Translation Learning)
Blended Learning Approach in English
Language Teaching – Its Benefits, Challenges,
and Perspectives
1 Introduction
Blended learning (BL) is nowadays a well-established methodology. The term itself was
firstly officially defined by Bonk and Graham [1] in 2006. They defined BL as learning
systems that combine face-to-face instruction with computer mediated instruction.
Generally, the BL approach aims to optimize student learning [2]. On the one hand,
there is more support and interactions online. On the other hand, it is expected that the BL
approach will enhance the quality of contact classes provided that students can benefit
from online learning activities and resources [2].
According to research studies [3–8], the BL methodology seems to be effective since
it brings several benefits for educational process, which are as follows:
• flexible learning and teaching (i.e., learners can access their learning materials from
anywhere and at any time and according to their own pace, and teachers can flexibly
modify students’ material, which they access online);
• improved pedagogy (it is assumed that well-thought use of new technologies
contributes to better use of relevant teaching methods);
• more frequent and timely feedback since feedback can be provided online;
• more learning resources, which are provided online;
• learners are expected to be more responsible for their learning, which might increase
their intrinsic motivation;
• collaboration among students and between students and teacher can be promoted more
often;
• use of new teaching methods and strategies enhanced by new technologies may
enhance learner performance;
• BL approach seems to be more cost-effective than the traditional, face-to-face
teaching.
Overall, the BL approach appears to be more effective than the use of only traditional
instruction. This was also confirmed by the findings of recent meta-analytic study by Vo
et al. [9]. Apart from the effectiveness of BL on learner performance in the institutions
of higher learning, they demonstrated that the effect of the BL approach is considerably
higher for the so-called STEM disciplines (i.e., hard disciplines such as chemistry or
public health) than for the so-called non-STEM disciplines (i.e., soft disciplines such
as English or psychology). Furthermore, Dziuban et al. [10] compared face-to-face
instruction, blended learning, and online learning and discovered that the BL approach
contributed to better learning outcomes than the fully online approach. In addition, in
some subjects the BL approach was more beneficial than the traditional one.
Blended
learning
Online
• eLearning
learning • Mobile
learning
Traditional
instruction
However, such an approach demands even more rigorous teaching methods and
strategies, as well as a more elaborate and meaningful context within which learning
can take place. In addition, the content itself must be tailored to learner needs in order
to motivate them to use a mobile application on a daily basis [29, 30].
296 B. Klímová and M. Pikhart
3 Conclusion
The findings of this overview article indicate that the emergence of new technologies
constantly influences the development of new learning modalities, which attempt to
respond to it. For further research into other aspects of the emergence of new technologies
and blended learning, business communication and eLearning see the research of Pikhart
[31–41]. At the moment, blended learning experiences a new phase of its development by
including mobile learning as valuable counterpart to traditional, face-to-face teaching.
This new BL approach, however, will need new challenging pedagogical methods in
order to enhance the effectiveness of the learning outcomes. Therefore, future research
should focus on the exploration of effectiveness of this new BL approach.
Acknowledgments. This study is supported by the IGS project 2019, run at the Faculty of
Informatics and Management, University of Hradec Kralove, Czech Republic.
References
1. Bonk, C.J., Graham, C.R.: The Handbook of Blended Learning Environments: Global
Perspectives, Local Designs. Joosey-Bass/Pheiffer, San Francisco (2006)
2. Kose, V.: A blended learning model supported with web 2.0 technologies. Procedia Soc.
Behav. Sci. 2(2), 2794–2802 (2010)
3. Khan, A.I., Noor-ul-Qayyum, Shaik, M.S., Ali, A.M., Bebi, C.V.: Blended learning process
in education context. MECS 9, 23–29 (2012)
4. Porter, W.W., Graham, C.R., Spring, K.A., Welch, K.R.: Blended learning in higher education:
institutional adoption and implementation. Comput. Educ. 75(6), 185–193 (2013)
5. Hubackova, S.: Blended learning – new stage in the foreign language teaching. Procedia Soc.
Behav. Sci. 197, 1957–1961 (2015)
6. Frydrychova Klimova, B.: Blended learning. In: Mendez Vilas, A., et al. (eds.) Research,
Reflections and Innovations in Integrating ICT in Education, pp. 705–708. FORMATEX,
Spain (2009)
7. Klimova, B.: Evaluation of the blended learning approach in the course of business English
– a case study. In: Huang, T.-C., Lau, R., Huang, Y.-M., Spaniol, M., Yuen, C.-H. (eds.) SETE
2017. LNCS, vol. 10676, pp. 326–335. Springer, Cham (2017). https://doi.org/10.1007/978-
3-319-71084-6_37
8. Klimova, B., Simonova, I., Poulova, P.: Blended learning in the university English courses:
case study. In: Cheung, Simon K.S., Kwok, L.-f., Ma, Will W.K., Lee, L.-K., Yang, H. (eds.)
ICBL 2017. LNCS, vol. 10309, pp. 53–64. Springer, Cham (2017). https://doi.org/10.1007/
978-3-319-59360-9_5
9. Vo, H.M., Zhu, C., Diep, N.A.: The effect of blended learning on student performance at
course-level in higher education: a meta-analysis. Stud. Educ. Eval. 53, 17–28 (2017)
10. Dziuban, C., Hartman, J., Moskal, P., Sorg, S., Truman, B.: Three ALN modalities: an institu-
tional perspective. In: Bourne„ J., Moore, J.C. (eds.) Elements of Quality Online Education:
Into the Mainstream, pp. 127–148 (2004)
11. Tomlinson, B.: Comments on Part 4. In: Tomlinson, B., Whittaker, C. (eds.)
Blended Learning in English Language Teaching: Course Design and Implementa-
tion (2013). https://www.teachingenglish.org.uk/sites/teacheng/files/pub_D057_Blended%
20learning_FINAL_WEB%20ONLY_v2.pdf
Blended Learning Approach in English Language Teaching 297
12. Siew-Eng, L., Muuk, M.A.: Blended learning in teaching secondary schools’ English: a
preparation for tertiary science education in Malaysia. Procedia Soc. Behav. Sci. 167, 293–300
(2015)
13. Tosun, S.: The effects of blended learning on EFL students’ vocabulary enhancement. Procedia
Soc. Behav. Sci. 199, 641–647 (2015)
14. Dudeney, G., Hoskly, N.: How to Teach English with Technology. Pearson Education Limited,
Harlow (2007)
15. Hubackova, S., Semradova, I.: Comparison of on-line teaching and face-to-face teaching.
Procedia Soc. Behav. Sci. 89, 445–449 (2013)
16. Simonova, I., Kostolanyova, K.: The blended learning concept: comparative study of two
universities. In: Cheung, Simon K.S., Kwok, L.-f., Shang, J., Wang, A., Kwan, R. (eds.)
ICBL 2016. LNCS, vol. 9757, pp. 302–311. Springer, Cham (2016). https://doi.org/10.1007/
978-3-319-41165-1_27
17. Frydrychova Klimova, B., Poulova, P.: Forms of instruction and students’ preferences - a
comparative study. In: Cheung, Simon K.S., Fong, J., Zhang, J., Kwan, R., Kwok, L.F. (eds.)
ICHL 2014. LNCS, vol. 8595, pp. 220–231. Springer, Cham (2014). https://doi.org/10.1007/
978-3-319-08961-4_21
18. Mauer, M.: Mobile Blended Learning with eSqirrel (2015). https://eproceedings.epublishing.
ekt.gr/index.php/openedu/article/viewFile/62/52
19. Elaish, M.M., Shuib, L., Ghani, N.A., Yadegaridehkordi, E., Alaa, M.: Mobile Learning
for English Language Acquisition: Taxonomy, Challenges, and Recommendations, vol. 5,
pp. 19033–19047 (2017). IEEE
20. Wu, Q.: Learning ESL vocabulary with smartphones. Procedia Soc. Behav. Sci. 143, 302–307
(2014)
21. Wu, Q.: Designing a smartphone app to teach English (L2) vocabulary. Comput. Educ. 85,
170–179 (2015). https://doi.org/10.1016/j.compedu.2015.02.013
22. Basal, A., Yilmaz, S., Tanriverdi, A., Sari, L.: Effectiveness of mobile applications in
vocabulary teaching. Contemp. Educ. Technol. 7(1), 47–59 (2016)
23. Mahdi, H.S.: Effectiveness of mobile devices on vocabulary learning: a meta-analysis. J.
Educ. Comput. Res. 56(1), 134–154 (2017)
24. Rezaei, A., Mai, N., Pesaranghader, A.: The effect of mobile applications on English
vocabulary acquisition (2014). https://www.researchgate.net/publication/261246911
25. Song, M., Chen, L.: A review on English vocabulary acquisition and teaching research in
recent 30 years in China. Sci. J. Educ. 5(4), 174–180 (2017)
26. Klimova, B.: Mobile phones and/or smartphones and their apps for teaching English as a
foreign language. Educ. Inf. Technol. 23(3), 1091–1099 (2017)
27. Klímová, B., Berger, A.: Evaluation of the use of mobile application in learning English
vocabulary and phrases – a case study. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R.
(eds.) SETE 2018. LNCS, vol. 11284, pp. 3–11. Springer, Cham (2018). https://doi.org/10.
1007/978-3-030-03580-8_1
28. Klimova, B., Prazak, P.: Evaluation of the effectiveness of the use of a mobile application on
students’ study achievements – a pilot study. In: Al-Sharhan, Salah A., Simintiras, Antonis
C., Dwivedi, Yogesh K., Janssen, M., Mäntymäki, M., Tahat, L., Moughrabi, I., Ali, Taher M.,
Rana, Nripendra P. (eds.) I3E 2018. LNCS, vol. 11195, pp. 37–44. Springer, Cham (2018).
https://doi.org/10.1007/978-3-030-02131-3_5
29. Wang, B.T.: Designing mobile apps for English vocabulary learning. Int. J. Inf. Educ. Technol.
7(4), 279–283 (2017)
30. Elaish, M.M., Shuib, L., Ghani, N.A., Yadegaridehkordi, E., Alaa, M.: Mobile Learning
for English Language Acquisition: Taxonomy, Challenges, and Recommendations, vol. 5,
pp. 19033–19047. IEEE (2017)
298 B. Klímová and M. Pikhart
31. Pikhart, M.: Communication based models of information transfer in modern manage-
ment - the use of mobile technologies in company communication. In: Proceedings of the
31st International Business Information Management Association Conference, pp. 447–450
(2018)
32. Pikhart, M.: Managerial communication and its changes in the global intercultural business
world. In: Special issue on education. SHS Web of Conferences, vol. 37 (2017)
33. Pikhart, M.: Intercultural linguistics as a new academic approach to communication. In: SHS
Web of Conferences, vol. 26 (2016)
34. Pikhart, M.: Theoretical foundations of intercultural business communication and their
practical consequences. In: SHS Web of Conferences, vol. 26 (2016)
35. Pikhart, M.: Current intercultural management strategies. The role of communication in
company efficiency development. In: Proceedings of the 8th European Conference on
Management, Leadership and Governance, pp. 327–331 (2012)
36. Pikhart, M.: Communication based models of information transfer in modern management
– the use of mobile technologies in company communication. Innov. Manage. Educ. Excell.
Through Vis. 2020, 447–450 (2018)
37. Pikhart, M.: Electronic managerial communication: new trends of intercultural business
communication. Innov. Manage. Educ. Excell. Through Vis. 2020, 714–717 (2018)
38. Pikhart, M.: Managerial communication and its changes in the global intercultural business
world. In: Web of Conferences, vol. 26 (2016)
39. Pikhart, M.: Intercultural linguistics as a new academic approach to communication. In: Web
of Conferences, vol. 26 (2016)
40. Pikhart, M.: Implementing new global business trends to intercultural business communi-
cation. In: Procedia Social and Behavioral Sciences, ERPA 2014, vol. 152, pp. 950–953
(2014)
41. Pikhart, M.: New horizons of intercultural communication: applied linguistics approach. In:
Procedia Social and Behavioral Sciences, ERPA 2014, vol. 152, pp. 954–957 (2014)
Investigating Students’ Use of a Social
Annotation Tool in an English for Science
and Technology Course
Jianqiu Tian(B)
Abstract. This study investigates the linguistic and pedagogical benefits and chal-
lenges of using a digital annotation tool (called Perusall) to facilitate second lan-
guage (L2) reading in an advanced English language course at university level. The
goals of the study are to analyze the students’ reading and annotating behavior,
examine the effects of social reading on their understanding of English for science
and technology texts and investigate how an L2 teacher might effectively incorpo-
rate this activity in his/her classroom. The results indicate that students spent an
above average amount of time reading (compared to that reported in the literature)
and that most students completed their reading assignments before class with the
Perusall platform. Moreover, they predominantly used the social reading expe-
rience to summarize the sections of the long article, and query fellow students
about the meaning of difficult and transitional sentences. Perusall allowed stu-
dents to co-construct meaning and scaffold their learning while engaged in close
readings of the science and technology texts outside of the physical classroom.
Drawbacks of social reading in this environment are primarily others’ comments
impeding some students’ understanding of the text and students’ frustrations with
some technical aspects of the Perusall tool. Pedagogical suggestions regarding L2
social reading include better integrating students’ virtual comments into classroom
discussion/activities, and offering more structure for students.
1 Introduction
Despite six years or more for English language learning, a majority of EFL students at
university level in China, especially students majoring in science and technology, are
overwhelmed by the English texts they are expected to read for their major studies. The
content and language integrated learning (CLIL) course on science and technology in this
research has been given to target the students’ reading ability, with reading assignments
of articles about 10 pages long each week. The limited class meeting time of two hours
a week rendered it necessary to adopt a “flipped” classroom format that requires the
students to read the texts before class and saves class time for more interactive activities
during which students can be actively engaged with instructors and other students to
consolidate their reading skills.
In this context, adequate reading before class is crucial to the success of the course. To
address the challenge of the students’ poor reading ability, Perusall, a digital annotation
tool (DAT), was deployed because in the digital environment the learners can not only
annotate and mark up a digital text [1–4], but can also share annotations with each other
on the same digital documents, and provide feedback on annotations [5, 6] to enhance
their reading comprehension.
DATs have been used in science and humanities courses, including language arts
courses, to ensure pre-class reading. In language courses the uses of DATs have primarily
been carried out with learners in L1 settings (e.g., [7–9]). The handful of studies using
DATs in L2 learning and teaching contexts are a language education course for pre-
service English-as-a-foreign-language (EFL) teachers [10], an English vocabulary and
reading course at university level [11], a course on a Spanish poetry [12] and a beginning
level Chinese language course [13]. This study is to be the first to report the use of DATs
in an EFL course with a focus on content of science and technology.
Thoms, Sung and Poole [13] investigates the linguistic and pedagogical benefits
and challenges of using a digital annotation tool (called eComma) to facilitate sec-
ond language (L2) reading in a second-semester, university-level Chinese language
course. The students primarily query fellow students about the meaning of vocabu-
lary/Chinese characters in the literary texts in the study, and eComma allowed students
to co-construct meaning and scaffold their close readings of the Chinese literary texts
outside of the physical classroom. The challenges encompass the students’ frustrations
with some technical aspects of the eComma tool and the instructor’s concerns about
integration of students’ social reading experiences outside of the class with in-class
discussions/activities. The authors make pedagogical suggestions of adding timing con-
straints to promote more virtual interaction, better integrating students’ virtual comments
into classroom discussion/activities, and offering more structure for novice learners.
3 Methodology
This research takes a social constructivist perspective on students’ pre-class collabo-
rative reading. Social constructivism suggests that students learn through the process
of sharing experiences and building knowledge and understanding through discussion
[15]. With this perspective, students in online learning communities [16] collabora-
tively build knowledge, verbalizing their thinking, building understanding, and solving
problems together [17, 18].
The specific research questions are: 1. How do students conduct pre-class reading on
Perusall, the social learning platform? 2.What is the efficacy of the platform in promoting
student learning? And 3. What are the challenges of using Perusall for learners?
3.2 Participants
Thirty-eight undergraduate students participated in the study, 28 males and 10 females,
ranging in age from 17 to 21 years old. Their native language is Mandarin Chinese. The
students brought a portable computer (laptop or tablet) and a smart phone to class to
accomplish in-class tasks. The students are in two classes, with twenty in class 1 (15
males and 5 females), and eighteen in class 2 (13 males and 5 females).
302 J. Tian
The DAT tool in this research is Perusall, an online social learning platform designed to
promote high pre-class reading compliance, engagement, and conceptual understanding
[19]. The instructor creates an online course on Perusall, uploading articles or documents,
and then creates reading assignments. Students asynchronously annotate the assigned
reading by posting (or replying to) comments or questions in a chat-like fashion.
An instructor view of the course home page is shown in Fig. 1. The instructor uploads
the reading material to the left-hand side of the page (under Documents) and then creates
specific reading assignments from these documents which appear in the right panel.
Figure 2 shows what a student sees after opening a reading assignment and high-
lighting a specific passage on a page in the assignment. A conversation window opens
on the right where the student can ask a question or make a comment.
Figure 3 shows a page that has been highlighted and annotated by students. When
a student clicks on a specific highlight that highlight turns purple, and the conversation
window for that highlight opens on the right.
When a student asks a question about a specific passage, it is automatically flagged
with an orange question mark, as shown in Fig. 3. Other students can respond in an
asynchronous conversation.
Perusall also has an integrated assessment tool that provides both students and
instructors with constant feedback on how students are engaging with the reading
assignments.
Moreover, Perusall provides many social features (sectioning, avatars, upvoting,
email notifications) that are designed to improve the interactions between students. Sec-
tioning allows the division of students into groups, which is suitable for class setting.
The avatars of other students and instructors who are viewing the same assignment at the
same time appear in the top left hand corner of the screen (Fig. 2), which increases the
Investigating Students’ Use of a Social Annotation Tool 303
social connectivity of the reading experience and encourages students to engage more
with the reading (through the software). Upvoting annotations provide feedback on the
annotations made by other students in their section. There are two types of upvoting in
Perusall, the orange question mark to solicit answers by other students and the green
checkmark to highlight a particularly helpful explanation. Finally, the email notification
feature alerts students not logged into Perusall when a classmate has responded to a
question or comment they have made.
Data collection took place during the Spring 2019 semester. The students were assigned
to read an article or two articles with a total length of about 10 A4 pages on a topic of
frontier of science and technology each week via Perusall for 14 weeks.
304 J. Tian
At the beginning of the semester, there was a guide for students on what to annotate,
including questions, opinions, strong pieces of evidence, key points, ideas with which
you disagree, good or poor supporting data or examples, inconsistencies, key terms or
definitions, contrasting points of view, key arguments, words with strong connotations,
and figures of speech (images that reveal the writer’s feelings). Then the students were
required to make at least 5 annotations on Perusall for each text before class and be
prepared for the reading tasks in class.
In particular, there was a crossover design for the topics Blockchains (week 12,
shortened as w12 hereafter) and Quantum Devices (w13), in which the students made
annotations as sections before class and then answered a set of specific or general ques-
tions alternately in class. Then the performance in the classroom task was compared
between the sections.
Another primary source of data for this study was comprised of students’ comments
and annotations on Perusall when reading articles on 14 topics across the semester.
Moreover, statistics were collected from the Perusall platform. There is also a survey of
the students’ behavior and perception of the benefits and challenges of Perusall.
40
Percent of students
30
20
10
0
<1 1-2 2-4 4-6 >6
The number of annotations students made ranges from 6.4 to 18.0 (Table 1).
As to the types of annotations, those of w12 blockchains include 37 questions, 241
comments and of w13 quantum consist of 25 questions, 243 comments.
The results indicate that students spent an above average amount of time reading
(compared to less than 3 h as reported in the literature [19]) and that most students were
active in making annotations before class with the Perusall platform.
Investigating Students’ Use of a Social Annotation Tool 305
Week W02 W03 W04 W05 W06 W07 W08 W09 W10 W12 W13 W14 W15
Total annotations 441 334 404 503 684 381 297 370 296 278 268 280 243
Average 11.6 8.8 10.6 13.2 18.0 10.0 7.8 9.7 7.8 7.3 7.1 7.4 6.4
What the Students Annotate. Content analysis of the students’ annotations for the w13
quantum text on the Perusall platform reveals the following types of annotations.
Summary of the text sections. An example reads as follows:
Quantum mechanics (at the atomic level, certainties-> probabilities, “su-
perposition, entanglement”) Applications: (1) Improving atomic clocks’ accu-
racy(entanglement) -> measuring tiny variations in gravity\spot underground
pipes\track submarines (2) Permitting messaging without worries about eavesdroppers.
(superposition, entanglement) (3) quantum computers (superposition, entanglement)
English explanation or Chinese translation of word meaning. Here are two examples:
Spookily adv. in an unusual and weird manner //Duplicate 复制
Comments showing the students’ own understanding of words, sentences, etc. A
student explains the sentence of “the odds are good; the goods, odd” as follows:
I deem that it has two meanings. First, it represent the odd of atoms of being
everywhere. Second it demonstrate the wide application of quantum technology. (sic)
Query and response about the meaning of difficult and transitional sentences. The
following example is a part of the original text followed by student query and response.
Born a century ago, this theory is the rule book for what happens at atomic scales,
providing explanations for everything from the layout of the periodic table to the zoo of
particles spraying out of atom-smashers. (original text)
Query: What’s the meaning of this? (this refers to the “zoo”)
Response: Maybe it refers to a place, situation or group marked by crowding,
confusion, or unrestrained behaviour, just like a big zoo with many lively animals.
Here in the text, zoo represents a huge number of unpredictable particles.
Explanation of the content from reference sources. Students also draw on reference
tools to gain further understanding of the content, such as:
Quantum mechanics, including quantum field theory, is a fundamental theory in
physics which describes nature at the smallest scales of energy levels of atoms and
subatomic particles. Classical physics, the physics existing before quantum mechanics,
describes nature at ordinary scale. -Wiki
Comments on the ideas in the text. The following paragraph is a student’s comment
on the current development of the quantum technology.
Actually I was shocked by the numbers of Chinese patents here. In my stereotyped
thoughts, US government/companies are devoting huge efforts in this (like Google), but
not much are done here (maybe I’m just lacking of information) (maybe our country
is trying to overtake in curves?). But the question appear, why China is so lagging
in developing quantum computing but have the same and prosperous development in
306 J. Tian
cryptography and sensors? Is it because the poor basis of the Chinese computer chips
and that actually delays the computing chip development in quantum?
It can be seen that the types of annotations are consistent with the particular diffi-
culties the part of the text involves or the students’ particular interests or understand-
ing. These annotations may help clarify misunderstandings or open space for further
discussion, which both potentially facilitate the students’ reading comprehension.
How the Student Use Perusall to Annotate. The survey indicates that most of the
students make annotations when they read for the second time after they figure out the
logic of the whole text in the first reading.
They may annotate thesis statements, other key sentences and long sentences that
defy understanding, the uncommon word, the understanding of the sentence or the asso-
ciation of social life. When they encounter difficulties, they look at Perusall to see
if fellow students have asked, or ask questions on Perusall. They also answer fellow
students’ questions.
After reading, students may write a summary, draw mindmaps, paraphrase some
sentences, or write a description of the chart.
These steps constitute a loop in which the students read the text in general, followed
by detailed reading, which subsequently resulted in enhanced understanding that was
consolidated in production tasks.
The question will be answered from two perspectives: the impact of the annotations on
the students’ understanding, and the students’ perceived benefits.
Relationship Between Student Reading Behavior and in-Class Performance. The
issues will be approached from the difference in answers to a question about under-
standing a particular sentence “The odds are good; the goods, odd” in the text of w13
quantum devices.
For this question, no student in Class 1 made any annotation of this point, but for
class 2, there was an attempt at explanation, followed by the instructor’s comment that
encourages other students to pay attention to the sentence, which resulted in an accurate
explanation.
The first student: The following are my understanding of the subtitle: “The odds are
good” means the counterintuitive quantum mechanics is very useful while “the goods,
odd” means that the products based on quantum mechanics were unexpected at first.
The instructor: Good attempt! Yet your explanation for the “odds are good” should
be one for “the goods, odd”, and an explanation is still pending for “the odds are good”.
The second student: “the odds” has the meaning of “the degree to which something
is likely to happen”, and “the odds are good” may mean that quantum technology is
very likely to have a promising future. (NB: This comment is upvoted by the instructor).
The in-class answers to the question indicate that among the 18 students in class 1
who answered the question, only two gave answers that were just close. On the other
hand, 5 out of the 17 students in class 2 (with annotations upvoted by the instructor)
who submitted an answer made a quite accurate answer. The difference in the in-class
Investigating Students’ Use of a Social Annotation Tool 307
task performance seems to suggest that the annotation facilitates the students’ attention
to and understanding of the part of the text.
Benefits Presented in the Survey. The following quotation is typical of the benefits of
Perusall students perceived.
Perusall does contribute to the understanding of reading materials both grammat-
ically and in terms of content, reducing some time for checking words and materials,
and strengthening the understanding of certain poorly understood paragraphs. A lot
of technical terms and principles would really have been very difficult for people who
don’t know much about them. The summative comments made by some students are also
extremely helpful.
In a word, the students not only benefited in understanding the literal meaning of
the texts, but achieved an enhanced comprehending the content. Moreover, they gained
a better overview of the text with the help of other students’ summative comments.
References
1. Abraham, L.B.: Computer-mediated glosses in second language reading comprehension and
vocabulary learning: a meta-analysis. Comput. Assisted Lang. Learn. 21(3), 199–226 (2008)
2. Ariew, R., Ercetin, G.: Exploring the potential of hypermedia annotations for second language
reading. Comput. Assisted Lang. Learn. 17(2), 237–259 (2004)
3. Johnson, T.E., Archibald, T.N., Tenenbaum, G.: Individual and team annotation effects on
students’ reading comprehension, critical thinking, and meta-cognitive skills. Comput. Hum.
Behav. 26(6), 1496–1507 (2010)
4. Kawasaki, Y., Sasaki, H., Yamaguchi, H., Yamaguchi, Y.: Effectiveness of highlighting as a
prompt in text reading on a computer monitor. In: Proceedings of the 8th WSEAS International
Conference on Multimedia systems and signal processing, pp. 311–315. World Scientific and
Engineering Academy and Society (WSEAS), Hangzhou, China (2008)
5. Cadiz, J.J., Gupta, A., Grudin, J.: Using Web annotations for asynchronous collaboration
around documents. In: Durand, D.G. (ed.) Proceedings of the 2000 ACM Conference on
Computer Supported Cooperative Work, pp. 309–318. ACM, (2000)
6. Jonassen, D.H.: Computers in the Classroom: Mindtools for Critical Thinking. Prentice-Hall,
Upper Saddle River (1996)
7. Gao, F.: A case study of using a social annotation tool to support collaboratively learning.
Internet High. Educ. 17, 76–83 (2013)
8. Lu, J., Deng, L.: Examining students’ use of online annotation tools in support of
argumentative reading. Australas. J. Educ. Technol. 29(2), 161–171 (2013)
9. Yang, X., Yu, S., Sun, Z.: The effect of collaborative annotation on Chinese reading level in
primary schools in China. Br. J. Edu. Technol. 44(1), 95–111 (2013)
10. Nor, N.F.M., Azman, H., Hamat, A.: Investigating students’ use of online annotation tool in
an online reading environment. 3L: Lang. Linguist. Lit. 19(3), 87–101 (2013)
11. Tseng, S.-S., Yeh, H.-C., Yang, S-h: Promoting different reading comprehension levels
through online annotations. Comput. Assisted Lang. Learn. 28(1), 41–57 (2015)
12. Thoms, J.J., Poole, F.: Investigating linguistic, literary, and social affordances of L2
collaborative reading. Lang. Learn. Technol. 21(2), 139–156 (2017)
13. Thoms, J.J., Sung, K.-Y., Poole, F.: Investigating the linguistic and pedagogical affordances
of an L2 open reading environment via eComma: An exploratory study in a Chinese language
course. System 69, 38–53 (2017)
14. Van Lier, L.: The Ecology and Semiotics of Language Learning: A Sociocultural Perspective.
Kluwer Academic Publishers, Boston (2004)
Investigating Students’ Use of a Social Annotation Tool 309
15. Vygotsky, L.S.: Mind in Society: The Development of Higher Psychological Processes.
Harvard University Press, Cambridge (1980)
16. Downes, S.: Creating an Online Learning Community [PowerPoint Slides]. https://www.
slideshare.net/Downes/creating-an-online-learning-community. Accessed 19 Aug 2019
17. Webb, N.M., Nemer, K., Chizhik, A., Sugrue, B.: Using Group Collaboration as a Window
into Students’ Cognitive Processes. National Center for Research on Evaluation Standards,
and Student Testing (CRESST), Los Angeles, CA (1995)
18. Crouch, C.H., Mazur, E.: Peer instruction: ten years of experience and results. Am. J. Phys.
69(9), 970–977 (2001)
19. Miller, K., Lukoff, B., King, G., Mazur, E.: Use of a Social Annotation Platform for Pre-Class
Reading Assignments in a Flipped Introductory Physics Class. Front. Educ. 3(8), 1–12 (2018)
The Application of Deep Learning in Automated
Essay Evaluation
Abstract. The shift from Automated Essay Scoring (AES) to Automated Essay
Evaluation (AEE) indicates that natural language processing (NLP) researchers
respond positively to the request from language teaching field. Writers and teachers
need more feedback about writing content and language use from AEE software
beside a precise evaluative score. This requirement can be met by the neural net-
work based deep learning technique. Deep learning has been applied in many NLP
fields and great success has been made, such as machine translation, emotional
analysis, question answering, and automatic summarization. Neural network based
deep learning is suitable for AES research and development since AES requires
mainly a precise score of writing quality. This can be accomplished with human
accurately scored essays as input and scoring model as output with deep learning
technology. However, AEE requires more than a score and deep learning can be
used to select linguistically meaningful features for writing quality and apply in
the AEE model construction. Related experiments already show the feasibility
and further research is worth exploring.
each composition and map them to the score of the composition. Here, the input end is
the composition text and the output end is the composition score. The neural network
can automatically construct the end-to-end algorithm from the composition text to the
composition score. The application of deep learning is not only the result of the devel-
opment of natural language processing (NLP), but also the historical necessity of the
development of NLP
(1) Construct a neural network with n layers and initialize network parameters.
(2) Input the untagged training data set into the first layer network, and input the results
of the first layer network into the second layer, and so on, until the nth layer network,
and output the results.
(3) Compare the network output results with the real results of the training data set,
and calculate the error. SGD, ADAM and other optimization methods are adopted
to modify the network parameters of all layers to reduce the error.
(4) Repeat steps (2) and (3) until the error meets the requirement.
(5) Accomplish the model for classification or prediction etc.
312 S. Ge and X. Chen
Among them, steps (2), (3), and (4) are called pre-training stage, which is an
important step to build a deep learning model (Fig. 1).
For the application and research of deep learning in different fields, the following
two problems should be solved first:
Machine translation refers to the process of translating one natural source language
into another natural target language by computer. In recent years, the performance of
Neural Machine Translation (NMT) has surpassed that of traditional Statistical Machine
Translation (SMT), and the research of NMT has become a new paradigm of machine
translation.
Kalchbrenner and Blunsom proposed for the first time a new end-to-end Encoder-
Decoder structure for machine translation [7]. The machine translation system generates
input sentences into vectors with Encoder and then generates the target language with
Decoder. At present, most of the NMT systems adopt this kind of Encoder-Decoder struc-
ture. When translating, the machine translation system first inputs the original sentences
of different lengths into the encoder, which converts them into fixed length vectors, and
then inputs the vector into the decoder, which can generate the translation of sentences.
The Application of Deep Learning in Automated Essay Evaluation 313
Sutskever et al. proposed the structure of Sequence to Sequence machine translation sys-
tem [8]. They tried to change the encoder and decoder into Long Short-Term Memory
(LSTM) structure. The experimental results show that this structure has more advantages
in accuracy than traditional SMT system.
However, since the encoder compresses all the information of the original sentence
into a fixed length vector, the longer the original sentence is, the more difficult the
system will be to process. In order to solve this problem, Bahdanau et al. first introduced
the attention mechanism into NMT system [9]. That is, after the encoder generates
a fixed-length vector, the vector is input into the attention layer, a weighted context
vector is generated, and then the vector is input into the decoder for decoding. This
structure enables the machine translation system to judge which word in the original text
is related to the word when it generates the word, which greatly improves the translation
performance when translating long sentences. Since then, attention mechanism has been
widely applied in NMT systems.
Gehring et al. showed that the performance of the NMT system using CNN as encoder
is similar to that of RNN, and the translation speed of the NMT system based on CNN
is faster than that of RNN under the same translation accuracy [10]. In the same year,
Google’s research team proposed a Transformer-based NMT framework, which aban-
doned the previous CNN or RNN structure and used a Self-Attention-based structure.
The framework can perform parallel operations on data during training, avoiding the
slow training speed caused by RNN’s need to read information sequentially. At the same
time, this architecture also enables NMT to achieve greater breakthroughs in translation
performance [11]. At present, more and more machine translation systems begin to adopt
a framework similar to Transformer.
Memory Network, which consists of a memory m and four components I (input feature
map), G (generalization), O (output feature map) and R (response) [14]. The network
framework has a great impact on the field of automatic question answering. Kumar et al.
proposed a framework of Dynamic Memory Network (DNN) [15]. This framework bor-
rows the structure of memory network and Decoder-Encoder, uses two GRU networks
to encode background information and questions respectively, and uses attention mech-
anism to encode questions and background information, and finally generates answers.
The framework achieves good results on the dataset of Facebook’s bAbI.
Neural network-based deep learning technology is becoming more and more mature,
and its application in automatic essay scoring is also worth exploring.
Alikaniotis et al. constructed a vocabulary representation model by designing an
algorithm to learn how certain vocabulary contributes to the essay score; LSTM network
was adopted to represent the meaning of the text; and an AES system was designed, which
achieved very good results [18]. LSTM network is a variant of RNN, and RNN is usually a
multi-layer structure. Multilayer neural networks can automatically learn useful features
from data, basic features from low-level structures and higher-level abstract features from
high-level structures [19].
On the basis of Collobert and Weston’s work [20], i.e., a distributed representation
of each word w in the local context of the corpus, and adopting the method proposed
by Tang [21], Alikaniotis et al. constructed a model that is extended to “capture not
only the local linguistic environment of each word, but also how each word contributes
to the overall score of the essay” [18]. They call this extended model the “augmented
The Application of Deep Learning in Automated Essay Evaluation 315
C&W model”. The core function of the model is to capture the “usage” information
of vocabulary, that is, to filter out the “under-informative” words, such as is, are, to, at
and so on. It focuses on the information-rich vocabulary, namely “score-specific word
embeddings” (SSWEs). By using SSWEs, continuous vocabulary representation for
each essay is obtained, and each essay is treated as a sequence of tokens to explore the
use of uni- and bi-directional LSTM networks, and finally these sequences of tokens
are embedded into fixed-length vectors. These word vectors representing the essays are
input into the linear units of the output layer, and the predicted scores of the essays are
obtained.
Alikaniotis et al.’s scoring model was used to evaluate English essays of middle
school students from Grade 7 to Grade 10 [18]. The essay dataset consists of 12,976
essays, ranging from 150 words to 550 words, all of which are scored manually to achieve
the final score for each essay. The experiment compares several models, and the final
results show that SSWE combined with two-layer bidirectional LSTM network achieves
the best results.
Another research on AES based on neural network is Taghipour and Ng [22]. In this
study, RNN is adopted to complete the training of essay scoring task and scoring model
construction.
Taghipour and Ng construct an automated essay scoring model with CNN archi-
tecture consisting of five layers: lookup table layer, convolution layer, recurrent layer,
mean over time, and linear layer with Sigmoid activation [22]. The first lookup layer
projects each word of an essay into a high-dimensional space. The convolution layer is
equivalent to a function, which extracts the possible local contextual dependencies from
the n-gram vectors transferred from the lookup layer, so as to improve the performance
of the scoring system. The recurrent layer first generates the embedding structure, then
processes the input data and generates the digital representation of a given essay. “This
representation should ideally encode all the information required for grading the essay.
However, since the essays are usually long, […] the learnt vector representation might
not be sufficient for accurate scoring”. Therefore, this study adopts and compares various
deep learning strategies, and finds that LSTM has the best effect. The mean over time
layer receives the processing results of the recurrent layer and calculates the average
vector of the same length for all the essays. The layer with Sigmoid activation maps the
vector into a scalar value, which is the score of the essay.
Taghipour and Ng [22] trained and evaluated the AES model based on deep learning
with the same essay dataset as Alikaniotis et al. [18]. The results show that the correlation
between the AES results of the model and the scores of the two human scorers is very
close, and the correlation coefficient is higher, but still a little lower than that between
the two human scorers. Since these two deep-learning-based AES studies are based on
the same dataset, a comparative analysis is also made. The Quadratic Weighted Kappa
(QWK) of Taghipour and Ng [22] is slightly higher than the best model of the latter
study [18].
The greatest advantage of the AES model or system based on the deep learning
algorithm of neural network is to avoid the heavy task of constructing evaluation features
manually. This method is worth exploring and applying if we consider only the accuracy
of essay score in large-scale examinations. However, another important function of
316 S. Ge and X. Chen
writing evaluation is to provide feedback for teachers and writers, that is, the highlights
and problems in writing texts. The deep learning method may improve the accuracy of
scoring, but from the writing theory, it is difficult to provide convincing arguments.
“The deep architecture of neural network models, however, makes it rather difficult
to identify and extract those properties of text that the network has identified as discrim-
inative” [18]. A big problem of the deep learning algorithm based on neural network is
that its overall operation is a “black box” structure, and it is usually difficult to find the
reason for the results, so it cannot give the feedback of the score.
Although Alikaniotis et al. tried to visualize the process of the neural network, it can
be clearly seen from the given example that the words marked as low quality and high
quality have little relationship with the quality of the essay [18].
In fact, in order to provide effective feedback for AES, deep learning can also be used
to mine the features of essay scoring.
Fu et al. believe that beautiful language is an important feature of students’ writing
performance, and it should play a certain role in essay scoring [23]. They propose a task
of elegant sentence recognition for essay evaluation, which mainly identifies elegant
sentences in middle school students’ Chinese essays for AES. The research presents a
deep neural network combining CNN and Bi-directional LSTM (BiLSTM) networks to
recognize grace sentences. The best result of the experiment is that the accuracy rate
of elegant sentence recognition is 89.23%. Using elegant sentence features in the AES
task can significantly improve the performance of the scoring system. Certainly, elegant
sentence has different definitions in different genres. Fu et al. define elegant sentences
in Chinese essays of high school students’ writing as “vivid language, flexible sentence
patterns, rhetorical structures, skillful borrowing and quotation, skillful use of classical
Chinese words, etc. [23] ” However, other genres, such as Business English, emphasize
“simple and clear” [24]. So, it is obvious that there are different definitions for elegant
sentences.
Language model has long been used in AES studies and research has proved that
RNN can build a very good language model [25, 26]. Kim et al. and Sundermeyer et al.
also carry out research of language modeling based on neural networks [27, 28].
Beside language model, syntactical analysis is also important in AES research.
Vinyals et al. adopt Sequence to Sequence model, regard the task of parsing as the
problem of sequence generation, and use LSTM network to obtain the syntactic tree [29].
On the one hand, the accurate annotation and extraction of these text features can
improve the accuracy of essay scoring, and more importantly on the other hand, it can
theoretically enhance the validity of AES, and provide efficient and accurate feedback
on language use for writers.
5 Summary
In recent years, thanks to the improvement of hardware computing ability, deep learning
has been widely used in NLP field. The algorithm model of deep learning has been
The Application of Deep Learning in Automated Essay Evaluation 317
improved day by day, and has made significant breakthroughs in machine translation
and other fields. The development of NLP technology will certainly promote the research
and application of AES which belongs to the application of this technology. However,
at present, there are still many problems to be solved:
(1) Deep learning learns the probabilistic features of language from corpus instead of
semantic features of language. Although deep learning has shown a strong ability
in NLP, there is still no perfect theory to explain the principle of deep learning,
which means the lack of theoretical support.
(2) Most of the research on NLP based on deep learning currently adopts data-driven
approach, and few studies can be closely integrated with linguistic research. In
order to enable machines to understand and analyze language in depth, future NLP
and linguistics, even cognitive science and other disciplines should promote each
other and make up for each other’s strengths and weaknesses.
Acknowledgements. This work is financially supported by the Science and Technology Project
of Guangdong Province, China (2017A020220002), Graduate Education Innovation Plan of
Guangdong Province (2018JGXM39) and the fund of Center for Translation Studies, Guangdong
University of Foreign Studies.
References
1. Whithaus, C.: Foreword. In: Shermis, M.D., Burstein, J. (eds.) Handbook of Automated Essay
Evaluation: Current Applications and New Directions, pp. vii–ix. Routledge, New York (2013)
2. Burstein, J.: Opportunities for natural language processing research in education. In: Gelbukh,
A. (ed.) CICLing 2009. LNCS, vol. 5449, pp. 6–27. Springer, Heidelberg (2009). https://doi.
org/10.1007/978-3-642-00382-0_2
3. Bengio, Y.: Learning deep architectures for AI. Found. Trends Mach. Learn. 2(1), 1–127
(2009)
4. Hinton, G.E., Osindero, S., Teh, Y.W.: A fast learning algorithm for deep belief nets. Neural
Comput. 18(7), 1527–1554 (2006)
5. Rosenblatt, F.: The perceptron: a probabilistic model for information storage and organization
in the brain. Psychol. Rev. 65, 386–408 (1958)
6. Xi, X., Zhou, G.D.: A survey on deep learning for natural language processing. Acta Automat.
Sinica 42(10), 1445–1465 (2016)
7. Kalchbrenner, N., Blunsom, P.: Recurrent continuous translation models. In: Proceedings of
the 2013 Conference on Empirical Methods in Natural Language Processing, pp. 1700–1709.
Seattle, Washington (2013)
8. Sutskever, I., Vinyals, O., Le, Q. V.: Sequence to sequence learning with neural networks. In:
arXiv:1409.3215v3 [cs.CL] (2014)
9. Bahdanau, D., Cho, K., Bengio, Y.: Neural machine translation by jointly learning to align
and translate. Proc. ICLR 2015, 1–15 (2014)
10. Gehring, J., Auli, M., Grangier, D., Dauphin, Y. N.: A convolutional encoder model for neural
machine translation. In: Proceedings of ACL, pp. 123–135 (2017)
11. Vaswani, A., et al.: Attention is all you need. In: Advances in Neural Information Processing
Systems, pp. 5998–6008 (2017)
318 S. Ge and X. Chen
12. Kim, Y.: Convolutional neural networks for sentence classification. In: Proceedings of the
2014 Conference on Empirical Methods in Natural Language Processing, pp. 1746–1751
(2014)
13. Wang, X., Liu, Y., Sun, C., et al.: Predicting polarities of tweets by composing word embed-
dings with long short-term memory. In: Proceedings of the 53rd Annual Meeting of Asso-
ciation for Computational Linguistics and the 7th International Joint Conference on Natural
Language Processing, pp. 1343–1353. Stroudsburg, PA (2015)
14. Weston, J., Chopra, S., Bordes, A.: Memory networks. In: Proceedings of ICLR 2015, pp. 1–15
(2015)
15. Kumar, A., Irsoy, O., Ondruska, P., Iyyer, M., Bradbury, J., Gulrajani, I., et al.: Ask me
anything: dynamic memory networks for natural language processing. In: Proceedings of the
33th International Conference on Machine Learning, pp. 1378–1387 (2015)
16. Rush, M. A., Chopra, S., Weston, J.: A neural attention model for abstractive sentence
summarization. In: Proceedings of EMNLP 2015, pp. 379–389 (2015)
17. Cao, Z., Li, W., Li, S., Wei, F.: Joint learning of focusing and summarization with neural
attention. In: Proceedings of 2016 COLING, pp. 547–556 (2016)
18. Alikaniotis, D., Yannakoudakis, H., Rei, M.: Automatic text scoring using neural networks.
In: Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics,
pp. 715–725. Berlin, Germany (2016)
19. Lee, H., Grosse, R., Ranganath, R., Ng, A.Y.: Convolutional deep belief networks for scalable
unsupervised learning of hierarchical representations. In: Proceedings of the 26th Annual
International Conference on Machine Learning, pp. 1–8 (2009)
20. Collobert, R., Weston, J.: A unified architecture for natural language processing: deep neu-
ral networks with multitask learning. In: Proceedings of the Twenty-Fifth international
conference on Machine Learning, pp. 160–167 (2008)
21. Tang, D.: Sentiment-specific representation learning for document-level sentiment analysis.
In: Proceedings of the Eighth ACM International Conference on Web Search and Data Mining
– WSDM 2015, pp. 447–452. Association for Computing Machinery (ACM) (2015)
22. Taghipour, K., Ng, H. T.: A neural approach to automated essay scoring. In: Proceedings of
the 2016 Conference on Empirical Methods in Natural Language Processing, pp. 1882–1891,
Austin, Texas (2016)
23. Fu, R., Wang, D., Wang, S., Hu, G., Liu, T.: Elegant sentence recognition for automated essay
scoring. J. Chin. Inf. Process. 32(6), 88–97 (2018)
24. Guffey, M.E., Loewy, D.: Essentials of Business Communication. Cengage Learning, South-
Western (2010)
25. Mikolov, T., Kombrink, S., Deoras, A., Burget, L., Cernocky, J.: RNNLM - recurrent
neural network language modeling toolkit. In: IEEE Automatic Speech Recognition and
Understanding Workshop (2011)
26. Chelba, C., et al.: One Billion Word Benchmark for Measuring Progress in Statistical
Language Modeling. Computer Science. arXiv:1312.3005 [cs.CL] (2013)
27. Kim, Y., Jernite, Y., Sontag, D., Rush, A.M.: Character-aware neural language models. In:
Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence. arXiv:1508.06615
[cs.CL] (2016)
28. Sundermeyer, M., Ney, H., Schlüter, R.: From feedforward to recurrent LSTM neural networks
for language modeling. IEEE/ACM Trans. Audio Speech Lang. Process. 23(3), 517–529
(2015)
29. Vinyals, O., Kaiser, L., Koo, T., Petrov, S., Sutskever, I., Hinton, G.: Grammar as a foreign
language. In: Advances in Neural Information Processing Systems. arXiv:1412.7449 [cs.CL]
(2015)
A Teaching Experiment
on a Knowledge-Network-Based Online
Translation Learning Platform
Hefei, China
Abstract. This paper aims to elaborate on the design and application of an online
platform as a knowledge-network-based system for online teaching/(self-)learning
of translation in/between English and Chinese. We have two purposes for this
research: first, to obtain a good understanding of translation trainees’ learning
behaviors in the corpus-assisted and knowledge-network-based translation learn-
ing setting, in the hope that sufficient data will be collected to draw a model
of knowledge-network-based learning. Equally important is our second purpose,
which is to initiate a more systematic and in-depth data-based empirical investiga-
tion into teaching designs for knowledge-based translation learning. This research
conducts an experiment on how teachers can use knowledge nodes to organize
online translation learning and how students perceive knowledge-network-based
learning. The experiment reveals a rising trend of students’ translation quality
and they generally hold a positive attitude towards this learning model. Based on
theoretical discussions of the platform design rationale and the findings from the
teaching experiment, this paper explores how the knowledge-network-based trans-
lation learning can assist students in forming more efficient translation learning
strategies.
1 Introduction
With the rapid development of technologies in education practice and research, language
teaching has also experienced tremendous progress in terms of technology-enhanced
modes. Translation teaching, an advanced form of bilingual teaching, has been faced
with challenges and opportunities of computer-assisted and data-based forms. In order
to improve the efficiency of translation learning and teaching, an online translation
teaching/learning platform, ClinkNotes Online Platform, has been designed and put into
use, which includes a knowledge-network-based system with annotations of translation
methods for the bilingual corpora and an automatic monitoring system for the tracking
of students’ learning records and historical performances.
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 319–328, 2020.
https://doi.org/10.1007/978-3-030-38778-5_35
320 Y. Mu and W. Yang
The knowledge base of this project is designed with interdisciplinary approaches with
recourse to computer science, knowledge engineering and management, translation stud-
ies, functional/text linguistics, language education, etc. to develop a groundbreaking and
cost-effective educational paradigm for the teaching/(self-)learning of English-Chinese
bilingual text-production in classroom/web-based settings to alleviate the pressure on
labor-intensive language/translation courses.
As the first attempt of its kind in the field, the cutting edge of this platform lies not in
the size of its databases but in its knowledge-based, theoretically-informed delicacy and
relevance of annotations and its teacher/learner-friendly data management. This plat-
form endeavors to build a knowledge management system based on the domain-specific
ontology for translation/bilingual writing, which features a computable network of inter-
related and hierarchically distributed conceptual representations of the knowledge in this
field, with “tag-words” as the knowledge nodes to form a roadmap of navigation and
also as the keywords to introduce theory-informed annotations (Fig. 1).
3.1 Participants
The participants were 23 sophomore students majoring in English. Before joining in the
specialized teaching experiment, they had already taken courses on translation between
Chinese and English for two semesters. Therefore, they have basic translation knowledge
322 Y. Mu and W. Yang
and skills. Also, they had been exposed to online learning before and was basically famil-
iar with online information technology. Each student was well instructed and informed
about the operation of the ClinkNotes Online Platform before learning through this
platform.
3.2 Procedures
Before the experiment, a Chinese-English translation test was conducted with the help
of a Chinese text of about 300 words to pre-test students’ translation ability. In the
experiment, students were instructed to learn 13 translation knowledge nodes related to
the text within a month through the ClinkNotes Online Platform. The knowledge nodes
are: (1) Parody; (2) Verb-Present Participle; (3) Idiom; (4) Alliteration; (5) Rhythm; (6)
TransferredEpithet; (7) Noun-Pronoun; (8) Echo; (9) Reduplication; (10) Onomatopoeia;
(11) Classifier; (12) Metaphor; (13) Intertextuality, covering five categories of translation
method, rhetoric, grammar, information distribution within sentences and cultural back-
ground knowledge. The knowledge nodes are correlated with other knowledge nodes to
form a knowledge network. After learning, students were again assigned to translate the
same text as a posttest (Fig. 2).
To ensure the reliability and validity of the experiment, students were not informed
in advance that the same text would be used for pretest and posttest, and they were also
stipulated not to use other electronic resources except the platform during the one-month
experiment, here hence to avoid students from referring to the reference translation after
the pretest. Students could refer to paper dictionary to complete their translation within
a certain limit of time. The test results were scored according to the scoring standards
for the translation part of TEM-8 (Test for English Majors, band 8, which is supposed to
be for senior students). The weighted scores were made by two teachers and averaged
as the final scores of the students being tested. The statistical software SPSS 19.0 was
used to compare the mean values of the test results to help understand the changes of
A Teaching Experiment 323
students’ scores before and after the experiment, so as to test the learning effect of the
teaching model.
The study also conducted a questionnaire survey after the test to understand students’
self-perception and recognition towards the translation teaching model of knowledge
network. The questionnaire was designed according to the Likert Scale, assigning 5
levels of point for different options with 5 the highest recognition and 1 the lowest. Also,
one-to-one interviews were made, during which the teacher would ask students questions
about the translation task, the platform and their learning experience. Combined all the
above explorations, the study then probed into the feasibility of this translation teaching
and learning model. The detailed procedures are shown in Fig. 3.
Online learning of
Pretest Posttest Online test
knowledge nodes
One-to-one
Score the translation Questionnaire Data collection
interview
According to the scores of the two teachers, the pretest and posttest score of 23 students
were calculated and counted. Assisted by SPSS 19.0, Q-Q plot was adopted to test the
normal distribution of the scores, the results are shown respectively in Figs. 4 and 5.
As can be seen from Fig. 4, most of the points are allocated around the straight
line and the scattered points are basically diagonally straight. Also, the points in the
detrended normal Q-Q plot of pretest randomly fall around the zero-scale line, so it can
be speculated that the pretest scores of the 23 students are approximately in line with the
normal distribution. When observed in the same way, the posttest scores can be taken as
normally distributed.
We then applied the paired sample t-test to explore the correlation and significance
between the test scores and the knowledge-network-based translation learning model
facilitated by SPSS 19.0. Here in this study, the hypothesis and standard should be
324 Y. Mu and W. Yang
firstly specified. Hypothesis H0 is that the changes between the two groups of data are
not correlated with this translation learning model, while that for H1 vice versa. Without
peculiar requirements, the standard α is set as 0.05 to decide whether to accept H0 or
not. The basic statistical information is shown in Table 1.
As can be observed from Table 1, the mean value of pretest scores is 70.08 and that
of posttest scores is 78.96. Combined with the characteristics of normal distribution of
students’ scores, it indicates that most students’ pretest scores are around 70 points while
for posttest 79, uprising of about 9 points when compared with pretest.
A Teaching Experiment 325
The paired samples correlations are suggested in Table 2 as above. It can be seen
from Table 2 that the correlation value is a positive value 0.507, and p = 0.014 < 0.05.
With the standard α = 0.05, hypothesis H0 is rejected but H1 accepted. That is, the
changes between the pretest and posttest scores are significantly correlated. A detailed
result of the paired samples test is shown in Table 3.
N Correlation Sig.
Pair 1 pretest & posttest 23 .507 .014
As can be seen in Table 3, the average difference of pretest and posttest score is
8.870. Meanwhile, the observed p = .000 < 0.05, suggesting that statistically H0 is
rejected while H1 accepted. Combined with the correlations between the two groups of
data, it can be stipulated that the knowledge-network-based translation leaning model
significantly helps in improving students’ posttest scores and the translation quality of
students’ posttest is improved when compared with that of their pretest.
A questionnaire survey and one-one-to-one interview were conducted after the test to
help further demonstrate the above statistical hypothesis. Totally 23 copies of question-
naire survey were sent out to students and 23 were effectively received with effective
rate 100%. According to the evaluation index of the five-level scale, if the option value
is between 1 to 2.5 points, it means that the students hold a negative attitude towards
the survey item; if 2.5 to 3.5 points, then neutral attitudes and if 3.5 to 5 points, then
positive attitudes. The detailed results of the survey items related to knowledge network
are listed in Table 4.
The standard deviation of all the survey items, except item 7, is less than 1.00 and
their overall number is small, suggesting a relatively low discrete degree in their data
distribution with the option of each students closer to the mean score. Also, the mean
326 Y. Mu and W. Yang
value of all the survey items listed is above 3.5 points, indicating their generally positive
attitude towards the listing items. Besides, for all the items, their scores are bigger than
1, showing no students holding a completely negative attitude.
A further probe into the specific survey items is made in combination of the one-to-
one interview. From students’ self-perception, it is generally believed that the 13 specific
knowledge nodes are positively correlated with the difficult points in translation (scor-
ing 4.09). They thought that “when I translate the texts again, I would consciously think
about which knowledge nodes could be applied to the text, and I found that some could
be directly used in the translation”, and thus the knowledge nodes are much helpful in
translating (scoring 3.91). Different students held different opinions on what type of
knowledge nodes is more helpful, but they generally agreed that many knowledge nodes
could help them deal with the difficulties they came across during their pretest, and
thereby improving their translation quality when compared with their previous trans-
lation (scoring 3.65). This can be well illustrated by their posttest scores, which is
consistent with their self-perception. As to the knowledge network interconnected and
formed by the 13 nodes, students’ acceptance level is relatively high (scoring 3.65).
They argued that this combination way of knowledge “let us intuitively understand the
connection among different knowledge nodes”, “I can master one node while also get
to know another”, “pretty systematic and overall”, “feel like they are in one system”,
“it much saves my time and energy while learning”, etc. Therefore, this way helps in
translation learning (scoring 3.87) and students approved of the method for translation
learning with the aid of the knowledge network (scoring 3.91).
It is worth noting that in view of the hypothesis put forward in item 7, the mean
value of scores is 3.48 points, a little bit lower than 3.5 points. Still, 5 people chose
the option scoring 2, and the standard deviation of this item is bigger than the other 7
items, indicating a variation in students’ options. A further interview targeting at this
item is made to students, especially those holding negative attitude. It is found that
their controversial points are mainly in the annotation of knowledge nodes and the
explanation of example sentences. Some believed that “it is the first time for me to learn
these knowledge nodes. But some nodes are way too professional to understand”, and
“some nodes are not easy for me to understand, and it becomes even more difficult for
me to understand when another node is involved in”. Some students held that “there are
repeated examples among different nodes. I understand that there may be several nodes
in the same example, but I am used to recalling knowledge through examples, and that
makes me a little confused.” The conflicting part is more concerned about the students,
as their learning habits and their knowledge base vary. But overall, students are willing
to cultivate this learning model in their future translation learning (scoring 4.09), as “the
learning efficiency is relatively high, and I can master several knowledge nodes at the
same time”.
It can be seen from the experiment that according to students’ self-perception, the 13
specific knowledge nodes are positively correlated with the difficult points in trans-
lation. The students generally accept the knowledge-network-based learning modes.
328 Y. Mu and W. Yang
Acknowledgements. The work was substantially supported by The National Social Science Fund
of China (Project No. 19BYY125).
References
1. Gao, J., Ding, K., Pan, Y., Yuan, J.: Analysis of the research status of knowledge network at
home and abroad. Inf. Stud. Theor. Appl. 38(9), 120–125 (2015)
2. Cai, B., Chen, G., Huang, X.: International research on knowledge network: present situation,
hotspot and trend — bibliometrics analyzing based on web of science. J. Xidian Univ. (Social
Science Edition). 27(4), 40–51 (2017)
3. Yang, X., Li, W., Gu, Y.: Research on the construction of knowledge element based teaching
resources organization model. Documentation, Inf. Knowl. 1, 101–107 (2016)
4. Yu, S., Duan, J., Cui, J.: A double spiral deep learning model based on learning cell platform.
Modern Distance Education Research. (6), 37–47 + 56 (2017)
5. Yan, Z., Li, M.: Technological pedagogical and content knowledge network — a new
knowledge framework for teachers in information age. China Educ. Technol. 4, 58–63 (2012)
6. Cao, L.: Studies on learning path and learning behavior characteristics in online learning.
Distance Educ. China 4, 25–30 (2014)
7. Yang, X., Li, J., Xing, B.: Behavioral patterns of knowledge construction in online cooperative
translation activities. Internet High. Educ. 36, 13–21 (2018)
8. Zhu, Y., Chen, J.: The theoretical conception and engineering practice of a corpus-based
computer-assisted translation. Technol. Enhanced Foreign Lang. Educ. 4, 52–57 (2015)
9. Zhu, C., Mu, Y.: Towards a textual accountability-driven mode of teaching and (self-) learn-
ing for translation and bilingual writing: with special reference to a CityU online teaching
platform. Chin. Translators J. 2, 56–62 (2013)
10. Mu, Y., Tian, L., Yang, W.: Towards a knowledge management model for online translation
learning. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R. (eds.) SETE 2018. LNCS, vol.
11284. Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03580-8_21
11. Tian, L., Mu, Y., Yang, W.: Designing a platform-facilitated and corpus-assisted translation
class. In: Hao, T., Chen, W., Xie, H., Nadee, W., Lau, R. (eds.) SETE 2018. LNCS, vol. 11284.
Springer, Cham (2018). https://doi.org/10.1007/978-3-030-03580-8_22
Design of Discipline Information System
for ‘Foreign Language and Literature’
Jing He(B)
1 Introduction
The core membership unit in academic systems is discipline-centered [1]. In order to
effectively promote the construction of world-class universities and first-class disci-
plines, the State Council of the PRC issued the “Overall Plan for Coordinating the Pro-
motion of World-Class Universities and First-Class Disciplines” on October 24, 2015.
The plan, also known as the “Double-First Class” initiative, aims to ultimately build
a number of world class universities and disciplines by the end of 2050, in an effort
to make China an international higher education power [2]. On September 21, 2017,
Chinese authorities released a selected list of universities, which will participate in the
country’s construction plan of world-class universities and first-class disciplines. 42
universities colleges will be developed into first-class educational institutions, and 95
universities will focus on building their preponderant disciplines into first-rate ones,
including the discipline of ‘Foreign Language and Literature’ from 6 universities [3].
‘Discipline construction’, generally refers to developing a discipline at a university, has
entered a period of great change, and its importance has become increasingly promi-
nent in China. It urges universities to set up an information system and management
mechanism to promote the development of each discipline on purpose.
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 329–334, 2020.
https://doi.org/10.1007/978-3-030-38778-5_36
330 J. He
Of the many who have designed a discipline information system, none has as yet
developed an information system especially for ‘Foreign Language and Literature’, and
failed to fully meet the needs of today’s users in discipline management. Li has devel-
oped a management information system based on VB.NET to look up fast corrective
information of disciplines with the help of internet [4]. This system includes six mod-
ules: teaching, scientific research, degree management, equipment management, library
data, and academic exchange. Wang has designed a web-based information system and
applied the date mining technology to the system to make it more intelligent [5]. The
system is of a three-tiered structure, which consists by a personal space, a school space
and a university space. Its functions include looking for and revising information of aca-
demic groups, papers, scientific research projects, academic monographs, and textbooks
related to disciplines. Chen has designed a management information system based on
ExtJS, consists of five modules: financial support, scientific research, discipline construc-
tion, maintenance and system help, the information includes teaching groups, scientific
research information, teaching conditions, and talent cultivation information [6]. Li has
designed an information management platform based on portal technology, aims to
integrate new applications with existing system [7]. The platform mainly carries out the
following functions: discipline information display and distribution, academic exchange,
tutor management, discipline information management (research interests, team groups,
and scientific achievements, teaching conditions, conferences and seminars), project
management, and interdisciplinary management. It also provides common functionality
such as content aggregation, single sign-on, personal customization, system integra-
tion, and full-text search. Gao et al. have designed a discipline information platform for
Shanghai University of Finance and Economics [8]. This platform carries out five func-
tions as data base management, discipline programming, discipline information display,
discipline evaluation, and system management. These researches mentioned above have
laid a foundation for the overall structure and main functional modules of the discipline
information system. However, all these information systems are designed at the back-
ground of the educational information in China, the main idea of these designs is to
save manpower, resources and time. Nowadays, discipline management has become the
primary purpose of the system.
resources, and environment. The connotation of discipline management requires that the
discipline information system conforms to the discipline rules, covers the whole course
of the discipline’s life cycle, and serves the integration and development of disciplines.
‘An information system (IS) is a set of interrelated components that together collects,
processes, stores, analyses, and disseminates data and information in an organization;
an information system provides a feedback mechanism to monitor and control its oper-
ation to make sure it continues to meet its goal and objectives’ [10]. Information system
has different types, such as personal IS, group IS, and enterprise IS [10]. A discipline
information system can be categorized as a group IS, which improves communications
and supports collaboration within a discipline. Based on a general model of an organi-
zation [10], this paper attempts to build an organizational model refined to disciplines
(see Fig. 1). Information system can support and work in the automated portions of an
organizational process of disciplines.
studies, foreign language education technology, country and regional studies, etc. [11].
Main features of the discipline are as follows: (i) the research object is divided into
two parts: linguistics and literature, which are parallel and relatively independent; (ii)
with the obvious characteristics of interdisciplinary and multidisciplinary, the discipline
itself is interdisciplinary and developing/integrating with other disciplines; (iii) taking
the advantages of languages, it is closely linked with international higher education
institutions and has rich international resources.
System Management. Manage users, permissions and default settings for this infor-
mation system, so that the functional departments and the responsibility departments
of the university can real-time view and manage the overall progress of the discipline
construction project.
Fig. 2. Modules of the discipline information system of ‘Foreign Language and Literature’.
5 Conclusion
Nowadays, ‘discipline construction’ has become a fundamental task of universities in
China, which includes the reorganization of resources, systems, groups and technologies,
the information technology requested is more complex and sophisticated, but there is
no comprehensive and all-purpose information system for disciplines yet, especially
for the discipline of ‘Foreign Language and Literature’. Therefore, the author gives the
334 J. He
References
1. Clark, B.R.: The Higher Education System: Academic Organization in Cross-National
Perspective, 1st edn. University of California Press, California (1986)
2. The State Council of the People’s Republic of China. http://www.gov.cn/zhengce/con-tent/
2015-11/05/content_10269.htm. Accessed 24 Oct 2015
3. Ministry of Education of the People’s Republic of China. http://www.moe.gov.cn/srcsite/A22/
moe_843/201709/t20170921_314942.html. Accessed 21 Sept 2017
4. Li, K.: Design and Construction of Information Management System of discipline construc-
tion, 1st edn. Nanchang University, Nanchang (2007)
5. Wang, X.: Analysis and Design of Subject Construction Management System Based on Web,
1st edn. Central South University, Changsha (2008)
6. Chen, R.: Design and Implementation of Discipline Management System based on Ext JS,
1st edn. Huazhong University of Science and Technology, Wuhan (2011)
7. Li, D.: Design and Implementation of Subject Information Management Platform Based on
Portal Technology, 1st edn. National University of Defense Technology, Hunan (2011)
8. Gao, L., Hu, Q., Wang, S.: Practice of discipline information platform in promoting discipline
construction. Chin. J. ICT Educ. 3, 73–76 (2014)
9. Zhu, M., Yang, X.: Management of disciplinary programs and development of modern
universities. J. Grad. Educ. 6, 12–18 (2011)
10. Stair, R.M., Reynolds, G.W.: Principles of Information Systems, 13th edn. Cengage Learning,
Mexico (2018)
11. The Sixth Discipline Review Group of the Academic Degree Committee of the State Council
of PR China. A Brief Introduction to Degree Awarding and Talents Training. 1st edn. Higher
Education Press, Beijing (2013)
12. China Academic Degrees Graduate Education information. http://www.cdgdc.edu.cn/
xwyyjsjyxx/xkpgjg/283494.shtml#3. Accessed 5 May 2019
AIE-TRST (Artificial Intelligence in
Education – Teacher’s Role for Student-
Centered Teaching)
Simplifying the Validation and Application
of Games with Simva
Abstract. The suitability of games for learning has been proven for many years.
However, effective application of games in education requires two important
stages: their initial validation, and their later use in the classroom. Serious games
should be validated prior to exploitation to prove their efficacy and usefulness
as tools for teachers, via larger experiments that include data collection, either
from in-game interactions or from external questionnaires; this, in turn, requires
dealing with data privacy regulations and informed consent. Once validated, seri-
ous games can then be applied in educational environments, where their effective
application is closely linked to the tools and preparation available to the teachers
and educators that use them. In this paper, we revise the steps and considerations
that need to be dealt with both when conducting experiments with games and,
later, when applying them as part of teaching in educational scenarios. For both
these stages, we provide guidance and recommendations to simplify stakeholders’
tasks, including the use of the tool Simva, which simplifies the management of
users, questionnaires, privacy, data collection, and storage.
1 Introduction
The application of Game-Based Learning (GBL) has greatly increased in the last years,
as many studies have proven the benefits of applying games in educational settings
[1]. The interactive nature of games increases the engagement of students in learning
activities, motivating them to progress and complete the in-game tasks [2]. This way,
students further improve their learning as a consequence of their gameplay. The benefits
of games, and in particular of their application in education, have attracted the attention
of many stakeholders: from researchers, game developers and designers trying to create
games that are effective tools for learning, to teachers, educators and institutions more
increasingly willing to apply games as part of their teaching activities. These stakeholders
are involved on different parts of the application of games for learning, and consequently
face very different issues when carrying out these experiments or applications with
games.
© Springer Nature Switzerland AG 2020
E. Popescu et al. (Eds.): SETE 2019, LNCS 11984, pp. 337–346, 2020.
https://doi.org/10.1007/978-3-030-38778-5_37
338 C. Alonso-Fernández et al.
On the one hand, researchers, game designers and developers are trying to promote the
application of games in education by conducting experiments to establish their effec-
tiveness and usefulness as a tool for teachers. For this and other purposes, experiments
usually include the collection of interactions from students’ gameplays. For instance,
the authors of [3] provide a practical guide of the use of games in experiments, including
the choice of game, event coding, data determination, participants and data collection.
These experiments applying games have multiple benefits but also have high costs in
terms of time and effort, both during preparation and their later execution. These issues
need to be dealt with by whoever oversees the application: commonly game design-
ers, game developers, or researchers. On research applications, these issues are dealt by
researchers themselves who do not tend to involve teachers or educators in the process.
This way, the researchers conducting these studies take an active role in the use of games,
preventing teachers from dealing with these issues. While this simplifies teachers’ tasks
on these applications, it can also complicate their work in the common case when they
are later going to apply the same games on their own.
These experiments generally include the collection of some in-game interaction data
from players. The field of Learning Analytics (LA) [4], which has greatly increased since
2011 [5], covers the collection and analysis of data from learning activities to understand
and improve learners’ processes and contexts. Building up from LA and focusing on
serious games, the field of Game Learning Analytics (GLA) extends this to the collection,
analysis and display of information on the activities and progress of player-learners. The
applications of GLA are wide and varied [6], including, among others, assessment and
student profiling. These applications can be used to validate game design, or to gain
insights that would otherwise be much harder to obtain [7].
On the other hand, teachers, educators and institutions need effective tools that simplify
the application of games in their classrooms. In fact, teachers still find it difficult to
integrate this learning approach into their regular practice [8], partly because real, long-
term applications of games, necessary need to be managed by teachers on their own with
only minimal external support. However, teachers or educators are generally not experts
in dealing with software or hardware requirements. Therefore, simplifying teachers’
tasks on these real-setting scenarios becomes a crucial step towards the advance of
game-based learning.
Authors have identified this issue and try to propose actions to help teachers adopt
games. For instance, the work of [9] presents a framework to model the process of
teachers’ adoption of games. According to this framework, teachers (1) become aware
of the innovation to be introduced, (2) focus on adoption and seek more information, (3)
engage in activities to measure pros/cons of the innovation and decide whether to include
it or not, (4) introduce the innovation and finally, (5) obtain feedback to reinforce their
choice. On this work, authors also pointed out several recommendations for teachers to
adequately adopt games as part of their teaching practice, including the appropriation of
the game by experiencing the activity before taking it to their students. They also found
Simplifying the Validation and Application of Games with Simva 339
out that rejection of the activity was motivated by fears and issues such as a perceived
lack of advantage compared to their previous teaching activity, misuse of the game, or
fear of losing control of their students.
The focus of our work is on educational videogames or serious games in general.
However, commercial videogames may also be used in education. The study of [10]
presents the advantages and disadvantages of using commercial videogames in exper-
iments. Among the advantages they point out, the following are especially significant:
ecological validity, lack of implementation times and/or external influence on the imple-
mentation, and reproducibility; while disadvantages include that modifications in the
games may be difficult or even impossible to conduct, the specificity of the hardware
used, and the difficulty of finding a game that is a good fit for a given set of purposes
and constraints.
On this paper we revise the considerations that need to be taken into account when
(1) conducting experiments with games in real educational scenarios and (2) applying
games as part of teaching. We provide guidelines for both researchers or game devel-
opers/designers and teachers for both scenarios. These guidelines are presented along
with the tool Simva, which simplifies some of the most costly parts of experiments and
game applications including questionnaires and data collection, storage, or participants’
management. The rest of this paper is structured as follows: Sect. 2 describes consid-
erations when conducting experiments, including the GDPR regulation and the use of
informed consents. Section 3 describes considerations for teachers when applying games
in their classes. Section 4 presents Simva and its features to simplify experiments for
both previous scenarios. Finally, Sect. 5 presents the conclusions of our work.
of participants while the activity is being carried out, this system should not hinder
privacy.
3. After the experiments: once the experiments are completed, some offline aggre-
gated information could provide information of interest for the game develop-
ers/designers or researchers. This feedback of the experiences could be provided
via aggregated visualizations or with some more complex techniques such as data
mining. This data analysis could also be simplified if the system that collects the data
does so in a standard format and allows for a user-friendly export of the data. If data
is going to be reused or maintained, it should also be defined (e.g. in the informed
consent).
2.1 GDPR
Before conducting the experiments, several requirements have to consider including
privacy, and legal regulations that may affect how data can and should be collected and
stored. These regulations will typically differ depending on the type of users participating
in the studies (e.g. minors, participants with intellectual disabilities) and the specific
characteristics of the studies, including the type of data to be collected, the collection
and storage system.
The new General Data Protection Regulation (GDPR) defines personal data as “any
information that relates to an identified or identifiable living individual” [12]. This
includes the scenarios where different pieces of information joint together can be related
to an individual as well as cases of using pseudo-anonymization. If individuals can not be
identified from some data, that data is no longer considered as personal data. It is impor-
tant to notice that GDPR protects all personal data collected regardless of the system
used to store the data (e.g. paper, computer) or to process it (manually or automatically).
such as minors or people with certain disabilities, informed consents can be collected
from their parents of legal guardians. Although some studies have debated whether
children have the capacity to give consent by themselves, the most widely accepted
and recommended option is that approval is given by some other responsible adult (e.g.
parents or teachers) [14].
The use of informed consents has been required in recent research such as the latest
projects of the European Commission. For FP7 projects (years 2007–2013), informed
consents were required when participants were minors, patients, immigrants or inca-
pacitated, or when the studies collected any personal data [15]. For the case of minors,
those guidelines requested the informed consent of parents or legal representatives, but
also the consent of children, with information sheets created according to the age of the
participants. For the H2020 program (years 2014–2020), guidelines stated the informa-
tion informed consents should include (aim of the research, methods, how data will be
collected, protected and if it is later going to be reused or destroyed) [16]. For children
or people unable to give consent (e.g. mental disabilities), consent is to be obtained from
their legally authorized representative.
Figure 1 details some of the issues informed consents should include, as a guideline
for researchers creating informed consents. Notice that this is not an exhaustive list,
depending on the study, some of this points will not be required (e.g. there may be no
benefits or risks) and others may be included. Their order is also optional.
adapted to their age so they can understand at least the purpose of the study they are
going to take part in.
The application of games in real scenarios by teachers or educators also includes dealing
with several issues at the different stages of the application. Among the steps that teachers
need to take into account, we include:
1. Before the application: first of all, teachers will have to choose an adequate game
that fits the curricula or provides a useful experience for their students. Before taking
the chosen game to students, it is recommended that teachers play the game so they
have the complete experience and fully understand the tasks that their students are
going to be asked to do. If there is any complementary material available (e.g. users’
guide), it is also highly recommended that teachers fully read and understand it as
it may provide additional context and information about the game and its goals and
design. Additionally, teachers may want to assess their students using some external
questionnaires. For these cases, the questionnaires should be defined and prepared
before the game is played and handed to students at the appropriate times (before
and/or after the application). The questionnaires may be handled on paper; on Sect. 4,
we present a simpler way to deal with questionnaires using Simva.
2. During the application: the previous preparation of teachers by playing the game
and/or reading any complementary material will simplify their tasks while students
are playing as they will be more aware of the steps they have to do. Additionally,
for teachers to keep control of the class and of their students’ progress, some way of
feedback or system displaying information would be highly welcomed. For instance,
if an Analytics System is receiving the in-game interaction data, teachers will be
able to keep track of what each student/player is doing (e.g. progress, actions, paths,
performance metrics). This will also allow teachers to perform interventions during
the gameplays: they may help students getting stuck or provide additional material
for advanced students.
3. After the application: if aggregated data is provided to teachers (either as global
visualizations or as aggregated metrics or reports), this information could be used
by teachers as means of players’ assessment. Depending on the game content and
purpose, a debriefing session may be recommended to revise the content included
in the game and even to relate it to the curricula to increase understanding and help
students transfer the knowledge learned in the game.
4 Simva
Simva is a tool designed to simplify the scientific validation of serious games [17],
as well as the assessment of students playing them, both tasks commonly measured
via comparison of pre-tests with post-tests. Simva has been already tested on differ-
ent scenarios to validate games, compare different game versions or carry out recall
experiments [18].
Simva includes many features that can help to simplify both researchers, game
designers or game developers’ tasks when conducting experiments with games on educa-
tional settings; and teachers’ tasks when applying games in their classes. As part of these
applications, Simva can help the different stakeholders to deal with issues including: stu-
dents’ management, students’ anonymization, control of access, level of completion, and
data storage and management. Details of how Simva helps to deal with those issues are
provided below.
Student Management. Simva works with classes of students, to which then provides
anonymization features and simplifies assigning questionnaires. Classes of students can
be created in Simva providing the number of students per class. The created classes are
then kept in Simva where questionnaires can be linked to classes.
Access Control. Games can be configured to require the anonymous token for players
to access the game. If so, the game then checks that there is a class created in Simva where
the introduced token is included. When games are configured to include questionnaires
in Simva, students will not be allowed to start a questionnaire unless their access token
is configured for that questionnaire. Additionally, players will not be able to access the
game until the questionnaire prior to the gameplay is completed. This check is also done
via Simva.
Level of Completion. While experiments are in play, the class view in Simva provides
additional information for stakeholders to keep track of players’ progress. On this class
view, Simva displays the questionnaires status for each player: started, finished or not
configured. This status is displayed for all questionnaires configured (currently a max-
imum of three questionnaires are available: a pre-test, a post-test and an additional
questionnaire). Simva also displays whether interaction data (traces) has been collected.
The top part of Fig. 2 displays an examples class in Simva where the three question-
naires are configured for the class. For each student (column “Code” on the left-part),
344 C. Alonso-Fernández et al.
Fig. 2. Simva screenshots: top part, class view depicting students’ anonymous tokens, question-
naires status and traces collected; bottom part, list of students with tokens to be cut and handed to
students.
we can see the status of all three questionnaires (“finished”, “started” or “not found”).
The right-most column provides the traces collected.
corresponds to. After the experiments have been completed, stakeholders can download
all collected data from Simva, automatically linked from each student together.
5 Conclusions
To promote the application of games in education, both of the major stages must be
simplified: first, the experiments carried out to validate and prove the efficacy of these
games as learning tools; and then, the teacher’s tasks when applying games on their
own. For the first stage, researchers or game designers and developers can benefit from
automated support that simplifies compliance with data privacy regulations such as the
GDPR, and the gathering of informed consent in experiments. In this paper we have
provided guidelines for both, and described the use of a tool to greatly simplify the
adoption of these guidelines through partial automation: Simva.
Once games are validated as effective, we enter a second stage, where teachers and
educators apply them effectively in their classrooms. This, again, requires tools and
preparation to manage the game application on their own. For this purpose, we have
provided guidelines on the steps that teachers should take before, during and after the
application of games. In our experience, these guidelines make teachers more comfort-
able with the application of games, making them aware of their students’ actions and
progress while they are playing, and providing them tools to conduct activities after the
game that can help students relate the content with the curricula.
Both stages can benefit from the use of Simva. Although the main goal of the tool
is to simplify the scientific validation of games, it can also help in everyday classroom
uses of serious games by teachers. Simva helps in the questionnaires’ management, data
collection, users’ management and privacy issues.
Future lines of work include testing Simva in more experiments, determining the
relative effectiveness of the tool for different stakeholders in each of their tasks, and iden-
tifying areas of improvement to further simplify the application of games in educational
scenarios.
References
1. Hainey, T., Connolly, T.M., Boyle, E.A., Wilson, A., Razak, A.: A systematic literature review
of games-based learning empirical evidence in primary education. Comput. Educ. 102, 202–
223 (2016)
2. Boeker, M., Andel, P., Vach, W., Frankenschmidt, A.: Game-based E-learning is more effec-
tive than a conventional instructional method: a randomized controlled trial with third-year
medical students. PLoS ONE 8, e82328 (2013)
3. Järvelä, S., Ekman, I., Kivikangas, J.M., Ravaja, N.: A practical guide to using digital games
as an experiment stimulus. Trans. Digit. Games Res. Assoc. 1, 85–115 (2014)
4. Long, P., Siemens, G.: Penetrating the fog: analytics in learning and education. Educ. Rev.
46, 31–40 (2011)
5. Long, P., Siemens, G., Gráinne, C., Gašević, D.: Proceedings of the 1st International Con-
ference on Learning Analytics and Knowledge, LAK 2011, 27 February – 1 March 2011,
Banff, Alberta, Canada, p. 195 (2011)
346 C. Alonso-Fernández et al.
Abstract. The 20th-century was the age of computers and information communi-
cation technology; at the beginning of the 21st-century researchers are exploring
the use of robots in the classroom. Our review investigates the implementation
of copresent social robots with teaching purposes in a classroom setting in areas
other than the teaching of subjects that are closely related to the field of Robotics.
We are interested in anthropomorphic robots, with an active role in the classroom
and capable of human-like activity. With a search of the WOS database and a sub-
sequent manual search in 19 journals we identified 24 relevant articles which have
been included in the analysis. Studies mostly include small number of participating
learners. In all studies special conditions are established for the robot interven-
tion in a classroom. Most often robots appear in roles as teacher, teacher assistant
and Care-Receiving Robot. Robots interventions were conducted by NAO, Saya,
RoboThespian, Bioloid, BAXTER, Darwin, NIMA-Robocop, Robosapien, TIRO.
Social robots diverge from the computer-mediated communication technologies,
as they are not mediating interaction but are partner in interaction. ITSs and ILEs
assist teachers in teaching, while the teacher and a robot have a shared presence
in the classroom. The copresent social robots perform a social role by interacting
with students. Robotic activities are aimed at delivering learning materials and not
primarily for individualised teaching, which encompasses the delivery of feedback
and the tailoring learning activities for individual learner’s needs.
1 Introduction
Intelligent tutoring systems (ITSs) and intelligent learning environments (ILEs) support
teaching and benefit learning outcomes [1]. By delivering instruction, ITSs and ILEs
support learning in diverse social modes for individual or group activities. ITSs and
ILEs supported individualised teaching are not limited to an individual student; it is also
conducted in small groups, large groups or peer learning. ITSs provide an individualised
learning experience in four main ways: monitor appropriateness of student’s input in
learning process, deliver appropriate complexity of task for students, provide effective
feedback based on the pedagogy model, apply interface/s for communication about the
domain or the learning contents [2].
This review study of robots in the classroom is focused on copresent robots as
physically embodied and physically present in the user’s space [3] and social robots
as »an autonomous, physically embodied robot that interacts and communicates with
humans by following social behaviours and rules attached to its role« [4 p. 628].
Social robots combine artificial intelligence and autonomous behaviours [5]. Social
robots diverge from the computer-mediated communication technologies as they are
not medium through which humans interact, but a medium with which humans interact
[6]. ITSs and ILEs assist teachers in teaching. While the robots’ behaviour and their
communication are connected with the social role that has been assigned to them [4].
They are hence capable of the appropriate verbal and non-verbal communication. Social
robots »can provide supportive behaviour, feedback and recommendations, as well as
attention acquisition to assist users in several applications« [7 p. 9].
“Social robots are being designed to deal with human care, health, domestic tasks,
entertainment and various other forms of immaterial and material tasks which aim to
renew human capacities” [8 p. 12], to take care of children, the elderly, the disabled
and the ill [8 p. 14]. They enter the reproduction sphere [8 p. 14]. Social robots are
being introduced as aides to the elderly [8–11], as durative assistants [10], in the field
of childcare [8, 11], at home [8, 12], in work environments and public spaces [12].
In the health care sphere [11–13] they are used for patients with dementia [11, 13],
patients with cognitive/motor disorders [11] and “in critical areas in medical care to
automate supervision, coaching, motivation, and companionship aspects of interactions
with vulnerable individuals” [11]. In the therapeutic domain [10, 12] they help with
autism [10, 11, 13, 14], they are used “as therapeutic tools for children, the elderly,
stroke patients, and other special-needs populations requiring personalised care” [15].
Short-term public interaction robots are used as visitors guides and as tour guides in
museums [10]. Social robots of the like of robot toys and robot pet companions have the
mission of engaging and entertaining the user [10].
Social robots are also used in education; most visibly for vocabulary learning [16] and
language learning [16, 17]. As they are capable of communicating and interacting with
students [5] and more specifically they are capable of delivering “a learning experience
through social interaction with learners” [18], they tend to be applied, tested and studied
in a variety of educational areas and purposes.
The introduction of social robots in different areas brings to the foreground the study
of human-robot communication [6, 19]. Researchers use them as test subjects, or more
precisely as research tools to »examine, validate and refine theories of social and biolog-
ical development, psychology, neurobiology, emotional and non-verbal communication,
and social interaction« [10 p. 20]. Robots could also serve as a tool for understanding
humans for example in language processing [17].
At the beginning of the 21st-century researchers are exploring the use of robots in
classrooms [14]. It is, more precisely a study of the technical capabilities of the robots
with teaching purposes and the Human-Robot Interaction (HRI) rather than research
into the pedagogical aspects of the interactions [20, 21].
The Robot in the Classroom: A Review of a Robot Role 349
From the reviews analysed it is clear that the majority of the performed studies focuses
on the utilisation of robots in the teaching of foreign languages [14, 17], robotics [14,
22] physics, mathematics [22], language, science and technology [17].
By analysing learning outcomes we see that they are predominantly cognitive [17]:
the understanding of concepts in the STEM fields, the development of reasoning,
problem-solving, social interaction and teamwork [22]. Studies in high education areas
explored the effects of robots on learning outcomes. Said studies focus more on affective
outcomes than on cognitive development [18]. However ultimately the effects of robot
interventions on learning outcomes are not sufficiently explored [19].
Some studies explore the underpinning learning theory behind robot interventions.
In these cases, constructivism is identified as predominant; they, however, lack evidence
of its integration in the pedagogical practice [19]. It has been found that, as the learning
design is not explored sufficiently, there is a lack of well-defined curriculum and learning
material for teachers [17]. The robot activities in the classrooms are also dependent on
the level of autonomy achieved by the robot [23] and on the role that has been assigned to
the robot. The review of the role of the social robot in the educational sphere highlighted
a variety of roles assigned. Among them are: the role of companion, collaborator [10],
peer [10, 17, 18], tutor [17, 18], tool [17], presenter, teaching assistant, teacher, novice
[18], Care-eliciting Companion [21]; the latter is classified by researchers in our papers
as Care-Receiving Robot (CRR).
Our review investigates the implementation of copresent social robots with teaching
purposes in a classroom setting in areas other than the teaching of subjects that are
closely related to the field of Robotics. We are interested in anthropomorphic robots,
with an active role in the classroom and capable of human-like activity. We examine the
utilisation of the chosen robot technology evident in the researched studies.
Copresent robots are physically embodied and physically present in the user’s space
[3]. The physical appearance of a robot is an important factor of HRI [10, 21, 24]. Robots
which have their own independent body tend to attract the users’ attention [25, 26]. It is
hence important that the robot’s morphology matches “its intended function” [10 p. 9].
Li (2015) highlights not only the robot’s physical embodiment but also the importance
of its physical presence in the physical space of the user. Copresent robots combine these
two factors. They are persuasive, receive attention and are perceived positively [3].
Social interactions are important in the learning process and cognitive development,
this is true also when the robot is present in the learning activity [17, 18]. Elements of
human social psychology regulate not only the social interactions between people but
also the social interactions between humans and robots [10, 24, 27]. The robot’s learning
activities can lead to positive educational outcomes only if the students maintain a
relationship with the robot [28]. This is in the longer term, a difficult task. It is therefore
important that the robot is capable of communicating with the human in a manner
appropriate to its role. The presence of the robot in the classroom tends initially, defined
as the first two weeks, to cause excitement. The excitement, transforms, in the third and
consequent weeks, into a stable interaction and satiation [29].
Long-term interaction capabilities are therefore very important when it comes to
the robots used in the classrooms; particularly if the robot is capable of speaking [29].
Considered the time component in the reviewed studies, we believe that it provides the
350 V. Rosanda and A. Istenic Starcic
first indicative information about the robotic capability of maintaining the attention of
the students and its ability to perform learning activities [20]. The research question of
our review study discussed in this paper is: How is robot role and robot type applied in
the reviewed studies?
2 Methods
The systematic literature review was conducted from November 2018 to January 2019.
Three stages were applied: (a) planning, (b) performing a search in the database and the
selection of relevant articles, (c) wrap up with reporting.
In the planning stage, the inclusion and exclusion criteria were defined for the review
study. In the planning phase, we decided to start the search in the WOS database and
afterwards, we identified the journals which were relevant for a further manual search.
The journals were selected based on: references in selected articles and weather the
included articles have been indexed in other databases.
The search stage started in November 2018 with the Web of Science (WOS) database.
The search terms applied were »robot*« AND »education« OR *learn*« OR »teach* «.
The journals with most articles in the WOS were selected.
Based on the results of the automatic search in the WOS database, the relevant jour-
nals were selected for further manual. The title and the abstract of the papers identified in
the automatic search were then read. The papers were then selected if the inclusion cri-
teria were met. Based on the study of references in the articles, the search also included
journals indexed in SCOPUS and IEEE databases. In November and December 2018
we continued with a manual selection within the 19 selected journals.
On the final stage of the search, the selected papers were analysed by reading the
whole article. At this stage, we included the papers that matched the following inclusion
criteria: the study about copresent social robots with teaching purposes in classroom
settings for other curriculum subjects than mechatronics and robotics. We excluded lit-
erature matching any of the following exclusion criteria: duplicated publications; only
one is kept [30]; studies that deal with the area of mechatronics, robotics learning activ-
ities, robot building, robot programming [22]; studies that deal with machine learning
(not related to human learning) [19]; the robot is used for healthcare training activities;
the robot is used for special education; the robot is used for therapeutic purpose, sick-
ness; researchers made use of zoomorphic robots; researchers made use of telepresence
robots, virtual robots or on-screen avatar; material other than scientific articles and arti-
cles which do not report on the intervention study. The inclusion criteria were meet by
24 studies, listed in references [11, 25, 26] and references from [31–49]. Papers marked
with an * contain two studies each.
The analysis stage took place from January to June 2019. Two papers are reporting
the findings. One paper focuses on the research outcomes, the research design, the
analysis of journals, educational levels and curriculum areas, participants structure and
learner-robot interaction (LRI) time [20]. This paper focuses on the robot role and robot
type.
The Robot in the Classroom: A Review of a Robot Role 351
3 Results
The analysis shows a clear increase in the number of research studies of the selected
robots being utilised in classrooms during the period 2013–2018 when compared to the
2006–2012 period [20]. The researchers have been addressing 6–12 years old age group
most often. This was followed by the 3–5 years old age group. They often designed
and performed the research with more than one educational level. None of the reviewed
studies dealt with the application of the chosen robot technology in the context of life-
long learning [20].
The focus of this analysis was the robot role. We analysed the roles that have been
assigned to the robot. As the role of the robot was not explicit in all of the analysed studies,
we categorised the role of the robot based on the context. In the categorization we used
the following roles: teacher [50], teaching assistant [51], tutor [52], CRR [36], peer [53].
In papers that were part of our review study, the robot was used as a teacher (29%),
teacher assistant (29%), CRR (17%), tutor (8%), peer (8%), classroom management
tool (4%) and tutor/CRR (4%) (Fig. 1).
Robot Role
35%
30%
25%
20%
15%
10%
5%
0%
In the reviewed studies, the robot-assisted teaching has been conducted on the deliv-
ery of a lesson. Lessons were mainly focused on the teaching of new content (17). Only
from five of these cases it is clear that the robot activities were conducted in the context
of the learning curricula, in two of them in kindergarten [11]. Only in one of all analysed
studies, the pedagogy is in the first plan: “teachers aligned robots with the curriculum-
based learning activity rather than the activity with the robot” [35]. Lesson delivery
was aimed at classrooms or groups of students. Regarding individualised teaching, in
a small number of studies the robot adapted further teaching materials for individual
student’s needs (6), more often (9) the robot purely provided feedback to students. We
can conclude that robot’s performance in terms of individualisation identified as the
personalisation of lesson progress is rather underestimated. The adaptation of the diffi-
culty level or the content to the student’s performance was identified only in 6 studies.
352 V. Rosanda and A. Istenic Starcic
Feedback is instead merely the adaptation of the robot’s behaviour based on the answers
given by the students.
Analysing the robot type, most of the studies (11) in this review worked with the robot
NAO, followed by the android Saya (4), RoboThespian (2) and Bioloid (2). All the other
robot types and more precisely BAXTER, DARwIn-OP (Darwin), NIMA-Robocop H21
version of NAO, Robosapien, TIRO, have been used only in one research (Fig. 2).
Robot Type
12
10
8
6
4
2
0
4 Discussion
The robotic teaching activities analysed included a small number of groups in individ-
ual research, often specially designed for the experiment and with a small number of
students, most often between the ages of 3 and 12 [20]. These activities are, as also indi-
cated by other studies, very rarely a part of the school curriculum [17]. In this context,
the robot-learner interaction often takes the form of one-on-one interaction. Therefore,
relatively little is known about the learning activities of the selected robotic technology
in a realistic context, as a group activity with multiple learners, or with group learning
dynamics approaching learning dynamics in a real classroom environment. Ultimately
relatively little is known about the learning activities of the selected robotic technology
in a real classroom environment [44]. As explained by Verner et al. (2016) “Pedagogical
literature provides recommendations for effective instruction, but their application to
robot-assisted learning has not yet been developed” [48]. Also, the timing and scale of
the robotic activities do not reflect the school’s learning realities. The studied robotic
activities took the form of a unique session or of a short-lived, fragmented interaction,
with students. This does not reflect school realities. The same reasoning and limitations
apply to the tasks performed by the robots. The classroom activities performed with
The Robot in the Classroom: A Review of a Robot Role 353
the help of robotic technologies were: short, well-defined and well-prepared. Once the
activities were implemented in the classrooms human assistance of varied intensity and
often diverse in its scope was also required. The scope of an intervention depended on
the robot autonomy level. Intense human involvement was required when the robot was
operating in a teleoperated or remotely controlled mode, and only minimal intervention
was required when it was performing in an autonomous mode [23].
In our review, teleoperated or remotely controlled robots were used in the longest
unique interaction sessions (30–60 min) [20]. The length and the continuity of the inter-
action sessions are ultimately determined by technical limitations [46], the problem of
maintaining long-term LRI [29], and the demanding lengthy preliminary preparation.
We believe that for all of these reasons, researchers are increasingly aiming to study
the dynamics of introducing robots in classrooms rather than focusing their research on
examining the effectiveness of robotic activities.
The robot teaching activities were focused on the following domains: science, tech-
nology and mathematics, English, geometry, computer science, sign language, subjects
of preschool age, stone-age items, mathematics tables, weekly spelling tests, geography
and sustainable development [20].
The most commonly used robot types were: NAO, Saya, RoboThespian and Bioloid.
The robots predominantly assumed the following roles when performing teaching
activities: teacher, teacher assistant and CRR.
During the teaching process, the robot teacher mostly performed only one of the
many tasks usually assigned to the human teacher. It was most commonly involved in
the teaching of new material (17 studies). The findings indicate that the teaching content
was largely out of alignment with respect to the regular curriculum. Only in one survey
“teachers aligned the robot with activity rather than the activity with the robot” [35].
The vast majority of the covered studies performed teaching activities outside the scope
of the regular curriculum.
While the advantage of using Intelligent tutoring systems is to individualise and tailor
the feedback to an individual learner, robotic activities are aimed at delivering learning
materials to either a group or an individual. Of the 24 studies, only 6 studies focused on
the individualisation in terms of adapting the content for further learning activities (6),
while a few (9) purely provided feedback to students.
We believe that defining the robots as performing the teacher role raises, given at the
current stage of development of robotic technology, expectations that robots cannot yet
meet. Among the open-ended questions, that need to be addressed, before the utilisation
of the chosen robot technology in everyday teaching practices, we list not only didactical
issues but also LRI [18, 46] and the educational component.
4.1 Conclusions
Social robots diverge from the computer-mediated communication technologies, as they
are not mediating interaction but are partner in interaction. ITSs and ILEs assist teacher
in teaching, while the teacher and a robot have a shared presence in the classroom. The
copresent social robot performs a social role by interacting with students.
Findings of our review study indicate that the research focuses mostly on studying
the human-robot interaction [20] in classroom practices. The research targets primarily
354 V. Rosanda and A. Istenic Starcic
objectives aimed at advancing robotic technology [21], and it does not necessarily focus
on pedagogical outcomes. Robotic activities are aimed at delivering learning materials
and not primarily for individualised teaching, which encompasses the delivery of feed-
back and the tailoring learning activities for individual learner’s needs. We conclude that
the chosen robot technology has not yet achieved a sufficient technical-developmental
phase, to allow for the focus of the research to shift to the educational-pedagogical
aspects of the introduction of the robot in the classrooms. The results of the review that
we have performed confirm findings of previous studies »the breakthrough of robots in
everyday teaching practice is not yet visible« [54 p. 293]. The development of educa-
tional robotic technologies is an interdisciplinary practice that urgently needs the input
of learning and educational theories. Exploring human learning for the advancement
of artificial intelligence and learning analytics will in turn effect in transformation of
human learning process [55].
Acknowledgments. The work of Andreja Istenič Starčič was financially supported by Slove-
nian Research Agency (P2-0210). This research has been conducted for “The 1st Workshop on
Artificial Intelligence in Education - Teacher’s Role for Student-centered Teaching - AIE-TRST
2019” organized by Andreja Istenic Starcic, Manolis Mavrikis, Maria Cutumisu, Cristina Alonso
Fernández.
References
1. Ainsworth, S., Grimshaw, S.: Evaluating the REDEEM authoring tool: can teachers create
effective learning environments? Int. J. Artif. Intell. Educ. 14(3) (2004). https://content.
iospress.com/articles/international-journal-of-artificial-intelligence-in-education/jai14-3-
4-03
2. Du Boulay, B.: Artificial intelligence as an effective classroom assistant. IEEE Intell. Syst.
31(6), 76–81 (2016). https://cutit.org/yN2nP
3. Li, J.: The benefit of being physically present: a survey of experimental works comparing
copresent robots, telepresent robots and virtual agents. Int. J. Hum.-Comput. Stud. 77, 23–37
(2015). https://cutit.org/QwFV8
4. Edwards, A., Edwards, C., Spence, P.R., Harris, C., Gambino, A.: Robots in the classroom:
differences in students’ perceptions of credibility and learning between “teacher as robot”
and “robot as teacher”. Comput. Hum. Behav. 65, 627–634 (2016). https://cutt.ly/JwdmtVu
5. Jung, S.E., Won, E.: Systematic review of research trends in robotics education for young
children. Sustainability 10(4), 905, 1–24 (2018). https://doi.org/10.3390/su10040905
6. Zhao, S.: Humanoid social robots as a medium of communication. New Media Soc. 8, 401–419
(2006). https://doi.org/10.1177/1461444806061951
7. Tsiakas, K., Kyrarini, M., Karkaletsis, V., Makedon, F., Korn, O.: A taxonomy in robot-
assisted training: current trends, needs and challenges. Technologies 6(119), 1–19 (2018).
https://doi.org/10.3390/technologies6040119
8. Taipale, S., de Luca, F., Sarrica, M., Fortunati, L.: Robot shift from industrial production to
social reproduction. In: Vincent, J., Taipale, S., Sapio, B., Lugano, G., Fortunati, L. (eds.)
Social Robots from a Human Perspective. Springer, Cham (2015). https://doi.org/10.1007/
978-3-319-15672-9_2. https://cutit.org/8Ziik
9. de Graaf, M.M.A., Allouch, S.B., Klamer, T.: Sharing a life with Harvey: exploring the
acceptance of and relationship-building with a social robot. Comput. Hum. Behav. 43, 1–14
(2015). https://cutit.org/rWPPg
The Robot in the Classroom: A Review of a Robot Role 355
10. Fong, T., Nourbakhsh, I., Dautenhahn, K.: Survey of Socially Interactive Robots. The Robotics
Institute Carnegie Mellon University, pp. 1–56 (2002). https://bitlylink.com/cmb3r
11. Fridin, M.: Storytelling by a kindergarten social assistive robot: a tool for constructive learning
in preschool education. Comput. Educ. 70, 53–64 (2014). https://doi.org/10.1016/j.compedu.
2013.07.043
12. Ahmad, M.I., Mubin, O., Orlando, J.: A systematic: review of adaptivity in human-
robot interaction. Multimodal Technol. Interaction, 3(14) (2017). https://doi.org/10.3390/
mti1030014
13. Kidd, C.D., Breazeal, C.: Effect of a robot on user perceptions. In: IEEE/RSJ International
Conference on Intelligent Robots and Systems (IROS) (2004). https://cutt.ly/XwdU0Ew
14. Cheng, Y.W., Sun, P.C., Chen, N.S.: The essential applications of educational robot: require-
ment analysis from the perspectives of experts, researchers and instructors. Comput. Educ.
126, 399–416 (2018). https://doi.org/10.1016/j.compedu.2018.07.020
15. Peca, A., Simut, R., Pintea, S., Costescu, C., Vanderborght, B.: How do typically developing
children and children with autism perceive different social robots? Comput. Hum. Behav. 41,
268–277 (2014). https://cutit.org/qrNMO
16. Shiomi, M., Kanda, T., Howley, I., Hayashi, K., Hagita, N.: Can a social robot stimulate
science curiosity in classrooms? Int. J. Soc. Robot. 7, 641–652 (2015). https://doi.org/10.
1007/s12369-015-0303-1
17. Mubin, O., Stevens, C.J., Shahid, S., Mahmud, A.A., Dong, J.J.: A review of the applicability
of robots in education. Technol. Educ. Learn. 1, 1–7 (2013). https://doi.org/10.2316/Journal.
209.2013.1.209-0015
18. Belpaeme, T., Kennedy, J., Ramachandran, A., Scassellati, B., Tanaka, F.: Social robots
for education: a review. Sci. Robot. 3(21), 1–9 (2018). https://doi.org/10.1126/scirobotics.
aat5954
19. Hong, N.W.W., Chew, E., Meng, J.W.S.: The review of educational robotics research and the
need for real-world interaction analysis. In: 2016 14th International Conference on Control,
Automation, Robotics and Vision (ICARCV), pp. 1–6 (2016). https://doi.org/10.1109/icarcv.
2016.7838707
20. Rosanda, V., Istenič Starčič, A.: A review of social robots in classrooms: emerging educational
technology and teacher education. Educ. Self Dev. 14(3), 1–20 (2019). https://doi.org/10.
26907/esd14.3.08
21. Sharkey, A.J.C.: Should we welcome robot teachers? Ethics Inf. Technol. 18, 283–297 (2016).
https://doi.org/10.1007/s10676-016-9387-z
22. Benitti, F.B.V.: Exploring the educational potential of robotics in schools: a systematic review.
Comput. Educ. 58, 978–988 (2012). https://doi.org/10.1016/j.compedu.2011.10.006
23. Beer, J.M., Fisk, A.D., Rogers, W.A.: Toward a framework for levels of robot autonomy in
human-robot interaction. J. Hum.-Robot Interact. 3(2), 74–99 (2014). https://doi.org/10.5898/
jhri.3.2.beer
24. Phillips, E., Ullman, D., de Graaf, M.M.A, Malle, B.F.: What does a robot look like?: a
multi-site examination of user expectations about robot appearance. In: Proceedings of the
Human Factors and Ergonomics Society Annual Meeting, vol. 61, pp. 1215–1219. SAGE
Publications, Thousand Oaks (2017). https://cutt.ly/1wdmjOs
25. Chin, K.Y., Wu, C.H., Hong, Z.W.: A humanoid robot as a teaching assistant for primary educa-
tion. In: IEEE Conferences, 2011 Fifth International Conference on Genetic and Evolutionary
Computing, pp. 21–24 (2011). https://cutt.ly/CwdmjVn
26. Chin, K.Y., Hong, Z.W., Chen, Y.L.: Impact of using an educational robot-based learning
system on students’ motivation in elementary education. IEEE Trans. Learn. Technol. 7(4),
333–345 (2014). https://cutt.ly/mwdmkjq
27. Mathur, M.B., Reichling, D.B.: Navigating a social world with robot partners: a quantitative
cartography of the Uncanny Valley. Cognition 146, 22–32 (2016). https://cutt.ly/hwdmk2b
356 V. Rosanda and A. Istenic Starcic
28. Kanda, T., Hirano, T., Eaton, D.: Interactive robots as social partners and peer tutors for
children: a field trial. Hum.-Comput. Interact. 19, 61–84 (2004). https://cutt.ly/HwdmlGF
29. Kanda, T., Sato, R., Saiwaki, N., Ishiguro, H.: A two-month field trial in an elementary school
for long-term human-robot interaction. IEEE Trans. Rob. 23(5), 962–971 (2007). https://doi.
org/10.1109/TRO.2007.904904
30. Spolaôr, N., Benitti, F.B.V.: Robotics applications grounded in learning theories on tertiary
education: a systematic review. Comput. Educ. 112, 97–107 (2017). https://doi.org/10.1016/
j.compedu.2017.05.001
31. Akalin, N., Uluer, P., Kose, H.: iSpy-uSign humanoid assisted interactive sign language tutor-
ing games. In: 2013 IEEE RO-MAN: The 22nd IEEE International Symposium on Robot and
Human Interactive Communication, pp. 290–291 (2013). https://cutt.ly/Ewdmhc3
32. Alemi, M., Meghdari, A.: The effect of employing humanoid robots for teaching english on
students’ anxiety and attitude. In: 2014 IEEE, Proceeding of the 2nd RSI/ISM International
Conference on Robotics and Mechatronics, pp. 754–759 (2014). https://cutt.ly/fwdmgJa
33. Baxter, P., Ashurst, E., Read, R., Kennedy, J., Belpaeme, T.: Robot education peers in a
situated primary school study: personalisation promotes child learning. PLOS ONE 12(5),
1–23 (2017). https://cutt.ly/YwdmgE9
34. Brown, L.N., Howard, A.M.: The positive effects of verbal encouragement in mathematics
education using a social robot. In: 4th IEEE Integrated STEM Education Conference (2014).
https://cutt.ly/Jwdmglu
35. Crompton, H., Gregory, K., Burke, D.: Humanoid robots supporting children’s learning in an
early childhood setting. Br. J. Educ. Technol. 49(5), 911–927 (2018). https://doi.org/10.1111/
bjet.12654
36. Ghosh, M., Tanaka, F.: The impact of different competence levels of care-receiving robot on
children. In: 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems,
pp. 2409–2415 (2011). https://cutt.ly/SwdmfVm
37. Han, J., Kim, D.: r-Learning services for elementary school students with a teaching assistant
robot. In: 2009 4th ACM/IEEE International Conference on Human-Robot Interaction (HRI),
pp. 255–256 (2009). https://cutt.ly/Dwdmfmc
38. *Hashimoto, T., Kato, N., Kobayashi, H.: Development of educational system with the android
robot SAYA and evaluation. Int. J. Adv. Robot. Syst. 8(3), 51–61 (2011). Special Issue
Assistive Robotics. https://journals.sagepub.com/doi/full/10.5772/10667
39. Hashimoto, T., Kobayashi, H., Kato, N.: Educational system with the android robot SAYA and
field trial. In: 2011 IEEE International Conference on Fuzzy Systems, pp. 766–771 (2011).
https://doi.org/10.1109/fuzzy.2011.6007430
40. *Hashimoto, T., Kobayashi, H., Polishuk, A., Verner, I.: Elementary science lesson delivered
by robot. In: 8th ACM/IEEE International Conference on Human-Robot Interaction (HRI),
pp. 133–134 (2013). https://cutt.ly/9wdmfqG
41. Keren, G., Fridin, M.: Kindergarten social assistive robot (KindSAR) for children’s geometric
thinking and metacognitive development in preschool education: a pilot study. Comput. Hum.
Behav. 35, 400–412 (2014). https://doi.org/10.1016/j.chb.2014.03.009
42. Llamas, C.F., Conde, M.A., Lera, F.J.R., Sedano, F.J.R., García, F.: May I teach you? Students’
behavior when lectured by robotic vs. human teachers. Comput. Hum. Behav. 80, 460–469
(2018). https://cutit.org/QgojI
43. Matsuzoe, S., Tanaka, F.: How smartly should robots behave?: comparative investigation on
the learning ability of a care-receiving robot. In: IEEE RO-MAN: The 21st IEEE International
Symposium on Robot and Human Interactive Communication, pp. 339–344 (2012). https://
cutit.org/PPKyH
44. Matsuzoe, S, Kuzuoka, H., Tanaka, F.: Learning english words with the aid of an autonomous
care-receiving robot in a children’s group activity. In: The 23rd IEEE International Symposium
on Robot and Human Interactive Communication 2014, pp. 802–807 (2014). https://cutit.org/
1S2lb
The Robot in the Classroom: A Review of a Robot Role 357
45. Pinto, A.H.M., Tozadore, D.C., Romero, R.A.F.: A question game for children aiming the
geometrical figures learning by using a humanoid robot. In: IEEE Conferences, 12th Latin
American Robotics Symposium and 2015 Third Brazilian Symposium on Robotics, pp. 228–
233 (2015). https://cutit.org/BwXq9
46. Serholt, S.: Breakdowns in children’s interactions with a robotic tutor: a longitudinal study.
Comput. Hum. Behav. 81, 250–264 (2018). https://doi.org/10.1016/j.chb.2017.12.030
47. Tanaka, F., Matsuzoe, S.: Children teach a care-receiving robot to promote their learning: field
experiments in a classroom for vocabulary learning. J. Hum.-Robot Interact. 78–95 (2012).
Inaugural Special Issue: Intersection of Systems Sciences and Human Sciences. https://cutt.
ly/IwdmzFT
48. Verner, I.M., Polishuk, A., Krayner, N.: Science class with RoboThespian using a robot teacher
to make science fun and engage students. IEEE Robot. Autom. Mag. 23, 74–80 (2016). https://
cutit.org/nyO83
49. You, Z.J., Shen, C.Y., Chang, C.W., Liu, B.J., Chen, G.D.: A robot as a teaching assistant in an
english class. In: Sixth IEEE International Conference on Advanced Learning Technologies
(ICALT 2006), pp. 87–91 (2006). https://doi.org/10.1109/icalt.2006.1652373
50. Podgoršek, S., Istenič Starčič, A., Kacjan, B.: The foreign teacher’s role in ICT supported
instruction. Sodobna pedagogika/J. Contemp. Educ. Stud. 70(136) (2019). https://cutt.ly/
Twdmw2S
51. Blatchford, P., Russell, A., Bassett, P., Brown, P., Martin, C.: The Role and Effects of Teach-
ing Assistants in English Primary Schools (Years 4 to 6) 2000–2003. Results from the Class
Size and Pupil Adult Ratios. (CSPAR) KS2 Project. School of Psychology and Human
Development, Institute of Education, University of London, pp. 1–37 (2004). https://cutt.
ly/rwdmcLD
52. Wood, D., Bruner, J.S., Ross, G.: The role of tutoring in problem solving. J. Child Psychol.
Psychiatry 17, 89–100 (1976)
53. Park, H.W., Kima, R.R., Rosenberg, M., Gordon, G., Breazeal, C.: Growing growth mindset
with a social robot peer. In: Proceedings of ACM SIGCHI, pp. 137–145 (2017). https://cutt.
ly/Xwdmp46
54. Johal, W., Castellano, G., Tanaka, F., Okita, S.: Robots for learning. Int. J. Soc. Robot. 10,
293–294 (2018). https://doi.org/10.1007/s12369-018-0481-8
55. Istenič, S.A.: Human learning and learning analytics in the age of artificial intelligence. Br.
J. Edu. Technol. 50(6), 1–3 (2019). Special issue editorial
ISTIL (‘I Search Therefore I Learn’)
Reordering Search Results to Support Learning
Bernardo.Nunes@anu.edu.au
Abstract. Although many learning activities involve search engines, their ranking
criteria are focused on providing factual rather than procedural information. In the
context of Searching as Learning, providing factual information may not be the best
approach. In this paper, we discuss the relevance criteria according to traditional
learning theories to support search engine results reordering based on content
suitability to learning purposes. We proceeded on the investigation by selecting
some self-proclaimed search literacy experts to answer thoroughly questions about
their views on the reordered results. We take into account that literacy expert’s
judgment may reveal issues regarded to technical side on learning supported by
search tools. Experienced users claimed a preference for reliable sources and direct
answers to what they are looking for, as they have exploratory skills to overcome
information incompleteness.
1 Introduction
Search engines are used to satisfy information needs. When inexperienced users1 search
for new information, they might not be skilled enough to deal with the complexity of
capturing, representing, matching and evaluating what they search for, what they need,
and what search engines provide them as answers. Then, the user may be affected by
a situation known as Anomalous State of Knowledge [4] in which insufficient previous
knowledge about the search subject interferes on his/her capability to formulate adequate
queries, impairing the results’ usefulness. According to Liu et al. [21], 47.19% of search
formulation are effective, which means that most initial attempts fail.
It is considered a hard task to properly build a suitable search query [4], and there
is also an obstacle to find relevant information within the high volume of retrieved
information. Since the number of Search Engine Result Pages (SERPs) increases based
on the number of documents that match the terms used, the order in which the content is
presented has an essential role. Results based only on matching search terms would rely
1 This study considers as inexperienced users as those with low search literacy or low knowledge
domain in the searched subject.
on pages based on reference indexes as relevance criteria. These types of criteria may be
useful to factual search, in which users look up for information performing navigational
or transactional searches tasks.
On the other hand, in procedural search users need to explore and to assemble
pieces of information to reach a comprehensive understanding and build knowledge.
Exploratory search [12] is defined as a set of tasks that goes beyond simple search, and
in which the user has to apply a greater cognitive effort to be able to investigate and learn
from the retrieved information.
Investigation and learning are also the focus of Searching as Learning (SaL) [15, 17],
for which the research agenda aims to address issues related to four main objectives: to
understand search as a human process; to measure learning performance and outcomes
during search; to establish relationship between learning process and search context;
and to design functionalities and search system interventions to promote learning [6].
Search engines should assume a more supportive role by presenting the content that
satisfies learning needs. Thus, they might be able to identify learning intent and reorder
indexed content based on this assumption. Page results order could satisfy learning
needs through instructional design principles. We consider cognitive relevance criteria
to reorder the retrieved information in a way to better attend educational aspects that
characterizes an exploratory search. From the SaL context, the focus of the current study
lies on how the information is presented to the user, in an attempt to customize results
to inexperienced users.
This paper is organized as follows: Sect. 2 presents learning theories in which this
study is grounded; Sect. 3 presents the related works; Sect. 4 presents the relevance
sorting criteria to reorder SERPs to support learning; Sect. 5 presents how the survey
and online interview were planned; Sect. 6 shows the results and discussions; and Sect. 7
presents the final remarks.
2 Learning Theories
This study is grounded on learning science. Learning does not necessarily imply in a
behavior change, however since an outcome is usually required for assessment, we take
into account the following statement: “Learning can be defined as changes in behav-
ior resulting from experience” [9]. Considering search engines as supporting tools to
learning, the experience cited by Lefrançois [9] comes from user’s interaction with
information. Two main approaches are predominant in this study: cognitivism and con-
structivism. While cognitive theories address issues of how information is received,
organized, stored and retrieved by the mind, constructive theories equate learning with
the creation of meaning from experience [1].
The value of what is learned is measured by how good it allows the learner to go
beyond the information already given [5]. Bruner [5] explained that learning derived
from the understanding of concepts and categories and from problem-solving proce-
dures. Besides that, his theory of discovery learning represents an important influence
on education, it means that the most effective way to construct knowledge is through
discovery. In this manner, search engines should provide conditions for it.
Reordering Search Results to Support Learning 363
3 Related Work
There are two steps required to provide an educational content ranking: (i) search query
classification to identify learning intent from query formulation and reformulation and
(ii) the arrangement of the retrieved content based on relevance criteria [19]. Educational
objects (search results) may be sorted through strategies within three categories [13]:
ranking based on text similarity; ranking based on user profile; and ranking based on
human review.
Since we do not present an automated solution yet, as text similarity review, we
consider a human review in order to investigate the main implications of this approach.
Human review approaches can be seen in solutions based on curatorship by specialists,
as it occurs in Learning Management Systems’ content sequencing [11] and knowledge
graph, with the goal of organizing relevant factual associations and entities [20].
However, neither a curriculum in educational platforms nor a knowledge graph are
completely available for any domain in a reliable manner. Therefore, we look for criteria
of sorting relevance to handle the SERPs and reorder them, for instance based on the
364 C. P. Teixeira et al.
We propose a set of relevance criteria to reorder the first page of search results based on
the relationship between concepts. This customized sorting reorder is based on the Smart
Insights2 research that indicates the importance of the first page as the most accessed of
SERPs. Hence, we considered that a new arrangement of the results based on different
content categories might be effective for learning purposes. As a teacher has the role to
motivate and clarify a subject by organizing the lecture in a flow that allows thoughtful
learning, we believe that somehow it could be reproduced by information systems when
assembling information pieces into logically chained and sorted content.
The proposal, represented in Fig. 1, is based on relevance sorting criteria that bring
the idea of content sequencing as a way to intentionally present a “learner friendly”
SERP and also based on learning of concepts and its relationships as Ausubel’s theory
of meaningful learning [2]. It is structured according to the following disposal:
2 https://www.smartinsights.com/search-engine-optimisation-seo/seo-analytics/comparison-of-
google-clickthrough-rates-by-position/.
Reordering Search Results to Support Learning 365
• Conceptual content: it starts with chained concepts in order to enhance the identifica-
tion of requirements, so that the learner does not access too complex content in the first
interaction with sources. Sometimes we need to understand how concepts are related
to each other in order to comprehend it. According to the concept of zone of proximal
development [18], a person develops a skill by maturing his/her awareness within a
particular internal context that includes the person’s actual level of knowledge about
the target domain, the types of help being offered and the sequence that these help
types are offered. The choice to show first conceptual concepts – especially content
which provides the user with an overview about the searched domain – is a way to
avoid a learned helplessness situation. Retrieve sequenced information showing the
relationship between the subject’s topics could be a guide to develop search engines
that are able to intervene in order to promote learning.
• Procedural content: allowing users to have access to guides and step-by-step material
aid them to be fully engaged with the content by providing tips on to how apply it.
Users can then organize the information based on what it relates to in order to create
the associations necessary to better understand through the general notion-application
relationship, consequently supporting learning and empowering them to be active in
the process by choosing directions to explore. This process is grounded on Ausubel’s
theory [2] of meaningful learning and intends to relate the retrieved online content
as meaningful signs, symbols and concepts that could be more easily incorporated
within the user’s cognitive structure. The choice to show procedural after conceptual
content is an attempt to promote the feeling of “all pieces fitting together”, in which
the contents read are meaningful and made sense. Focusing on applied concept as a
second level of organization has the purpose to present the information as useful and
not limited to memorization but as a way to build a conceptual framework regarding
how the searched subject can be interpreted and applied by the user.
• Deepened content: results capable of fetching specific information from what was
searched (in an implementation, it may be used specialization of query reformulations
from searches with similar topics). Based on Piaget [14] and Vygotsky [18] theories,
the user will handle more complex content after the previous steps, in a process
similar to constructivism’s assimilation. As users assimilate new information, they
can incorporate it into an already existing framework (provided by the conceptual
content) without changing that framework (incorporating the deepened content into
the conceptual framework about the searched subject).
• Related content: aligned to Bruner’s assumption [5], the user needs to go beyond
the information given. Here a variation of the spiral curriculum approach is applied,
in which each subject is revisited at intervals, with a more sophisticated level being
focused each time. Instead of deepen the subject; the idea is to retrieve related infor-
mation about the searched subject to provide a better understand on how they mold
together. The proposed approach aims to reveal concept chaining in a wider perspec-
tive instead of requiring a more focused search to avoid duplicate content. One goal
is to reduce the effort to identify related concepts that is required to increase users’
comprehension, by means of assisting users toward possible exploration paths and
serving as a guide to query reformulation. It represents an educational view that could
366 C. P. Teixeira et al.
be applied whenever is identified a learning intent from the user. It may also repre-
sent an attempt to present the content in a flow that could contribute to motivation
reinforcement.
5 Survey Setup
The success of information systems is influenced by the intent to use it; thus, the user’s
opinions are important [7]. To understand users’ point of view, a qualitative investigation
was conducted to inquiry a group of participants, which were selected based on their
search literacy skills. We take in account that a search literacy expert’s judgment may
reveal issues regarded to technical side on learning supported by search tools. Seven
experienced users were volunteers to an online interview, based on the following script:
1. How do you describe your searching and learning skills (i.e.: search literacy) and
what makes you good at it? – We want to identify which skills a customized search
engine could provide.
2. What kind of search strategies do you use to learn on the internet? – This ques-
tion intends to map suitable approaches used by experienced users to increase their
awareness about a subject.
3. While searching for the information what kind of problems do you face when you
have to learn something new? – We want to know which challenges an experienced
user face.
4. Comparing the following images of search engine results pages, which pages you
would visit, which pages you would not visit and why. – This question seeks to
provide a better understanding about results’ usefulness through the indication of
the pages that would be accessed or not, and to explain why.
The images of SERPs of the fourth question represented procedural search. The
search subjects were chosen based on their classification3 of most asked questions on
Google (88th and 131st respectively): “how to start a business” and “how to create a
website”. These subjects represent a type of questions that do not match simple answers,
it requires some information-literate actions in order to satisfy (learning) user’s intent.
The results of the first page were reordered according to the relevance criteria presented
in the previous section.
The seven users chosen to be interviewed were selected from the group of people who
claimed to have excellent search literacy skills. Their answers brought some insights
regarding what could be expected from a learning driven search engine and also to the
applied relevance criteria:
3 https://www.mondovo.com/.
Reordering Search Results to Support Learning 367
• Experienced users can properly formulate queries to find online whatever they are
looking for from the first search query used. It indicates that even the first queries
at the beginning of a search session can indicate if the user is an experienced or
inexperienced one. The recognition of a user’s struggle to formulate a query might
take place in this early search session timeframe by mapping the user’s behavior to
identify searching behavior patterns. An example of such mapping was handled by
[16], in which the user’s search behavior is seen as exploratory and modeled as a
Knowledge-intensive Process (KiP).
• Experienced users follow strategies like focused search in academic repositories and
reliable sources and semantic search. This kind of user’s decision-making also may
help to distinguish a novice from an expert user. An aggregated search, as described
in [3] could help novice users to envision wider possibilities and properly adopt the
best strategy.
• The main problems any kind of users face concern to the type (documents not related
to education) and content depth (superficial or incomplete). As a reflex of the huge
number of documents on the web, novice searchers take the risk of not finding the
most suitable document for learning purposes. As the most suitable document may
vary according to the user’s intent, the challenge is to identify the intent first rather
than to tune an algorithm’s document-query matching.
• Users worry more about sources than content. Reliability is pivotal and also should
direct educational content indexing. Categories of information and content usefulness
based on source reliability can be used to organize the documents, as indicated by [3]
and [10]. According to participants’ impressions the awareness of categories of the
results would make the approach more acceptable.
Although the above topics could help to detect learning needs that search engines
do not assist properly; to promote strategies to intervene in a Searching as a Learning
process; and to highlight challenges faced by SaL agenda, still remains a lack of formal
models to SaL and a lack of grounded theories that could better connect search process
to the learning sciences.
7 Conclusions
The understanding of users’ acceptance of SERPs sorting reorder and further analy-
sis about the reasons why a document link is considered useful will help to improve
our understanding about the requirements needed to customize a search engine able to
support educational searches. It was the major motivation behind this research and we
took into account search literacy user’s opinion in order to reveal critical issues on the
perspective of the search tool that even an expert could not overcome to satisfy his/her
learning needs.
As the current search engines sorting relevance criteria of referenced page links,
users search history and bookmarks and document-query matching does not necessarily
improve results suitability for learning purposes (e.g. a professor’s blog might be more
useful than an article from The New York Times), an alternative strategy to reorganize
SERPs, such as the one presented in this paper, is a necessary step. Using the searching
368 C. P. Teixeira et al.
experience of skilled users seems a reasonable alternative to aid novice searchers while
performing searches with learning intent.
This work was limited to aspects related to SERPs sorting reorder; however, we
could notice how important are users’ behavior and interaction with sources to advance
SaL research agenda. As future work, we envision:
Acknowledgments. This study was financed in part by the ‘National Council for Scientific
and Technological Development (CNPq) - Brazil’ - Process 315374/2018-7, Project ‘Search-
ing as Learning: the information search as a tool for learning’ and by the ‘Coordination for the
Improvement of Higher Education Personnel’ (CAPES) – Brazil – Finance Code 001.
References
1. Ertmer, P.A., Newby, T.J.: Behaviorism, cognitivism, constructivism: comparing critical
features from an instructional design perspective. Perform. Improv. Q. 6(4), 50–72 (1993)
2. Ausubel, D., Novak, J., Hanesian, H.: Educational Psychology: A Cognitive View, 2nd edn.
Holt, Rinehart & Winston, New York (1978)
3. Kopliku, A., Pinel-Sauvagnat, K., Boughanem, M.: Aggregated search: a new information
retrieval paradigm. ACM Comput. Surv. 46(3) (2014). 31 pages, Article 41. https://doi.org/
10.1145/2523817
4. Belkin, N.J., Oddy, R.N., Brooks, H.M.: Ask for information retrieval: part I. Background
and theory. J. Doc. 38(2), 61–71 (1982)
5. Bruner, J.S.: The narrative construction of reality. Crit. Inq. 18(1), 1–21 (1991)
6. Collins-Thompson, K., Hansen, P., Hauff, C.: Search as learning. Report from Dagstuhl
Seminar 17092 (2017)
7. DeLone, W.H., McLean, E.R.: Information systems success: the quest for the dependent
variable. Inf. Syst. Res. 3(1), 60–95 (1992)
8. Smith, C.L., Rieh, S.Y.: Knowledge-context in search systems: toward information-literate
actions. In: Proceedings of the ACM SIGIR Conference on Human Information Interaction
& Retrieval (CHIIR 2019), pp. 55–62 (2019)
9. Lefrançois, G.R.: Theories of Human Learning: What the Professor Said, 6th edn. Cengage
Learning, Wadsworth (2011)
Reordering Search Results to Support Learning 369
10. Huurdeman, H.C., Wilson, M.L., Kamps, J.: Active and passive utility of search interface
features in different information seeking task stages. In: Proceedings of the 2016 ACM on
Conference on Human Information Interaction and Retrieval (CHIIR 2016), pp. 3–12. ACM,
New York, NY, USA (2016)
11. Manrique, R.: Automatic learning content sequence via linked open data. In: International
Semantic Web Conference (DC@ ISWC 2017) (2017)
12. Marchionini, G.: Exploratory search: from finding to understanding. Commun. ACM 49(4),
41–46 (2006)
13. Ochoa, X., Duval, E.: Relevance ranking metrics for learning objects. IEEE Trans. Learn.
Technol. 1(1), 34–48 (2008)
14. Piaget, J.A.: Equilibração das estruturas cognitivas: problema central do desenvolvimento.
Trad. Álvaro Cabral. Rio de Janeiro, Zahar (1976)
15. Rieh, S.Y., Collins-Thompson, K., Hansen, P., Lee, H.J.: Towards searching as a learning
process: a review of current perspectives and future directions. J. Inf. Sci. 42(1), 19–34
(2016)
16. Tibau, M., Siqueira, S.W.M., Nunes, B.P., Bortoluzzi, M., Marenzi, I.: Modeling exploratory
search as a knowledge-intensive process. In: 2018 IEEE 18th International Conference on
Advanced Learning Technologies (ICALT), Mumbai, pp. 34–38 (2018)
17. Vakkari, P.: Searching as learning: a systematization based on literature. J. Inf. Sci. 42(1),
7–18 (2016)
18. Vygotsky, L.S.: Mind in Society: The Development of Higher Psychological Processes. Har-
vard University Press, Cambridge (1978). (Edited by Cole, M., John-Steiner, V., Scribner, S.,
Souberman, E.)
19. Yilmaz, T., Ozcan, R., Altingovde, I.S., Ulusoy, Ö.: Improving educational web search for
question-like queries through subject classification. Inf. Process. Manage. 56(1), 228–246
(2019)
20. Voskarides, N., et al.: Weakly-supervised contextualization of knowledge graph facts. In:
The 41st International ACM SIGIR Conference on Research & Development in Information
Retrieval (SIGIR 2018), pp. 765–774. ACM, New York, NY, USA (2018)
21. Liu, C., Gwizdka, J., Liu, J., Xu, T., Belkin, N.J.: Analysis and evaluation of query refor-
mulations in different task types. In: Proceedings of the American Society for Information
Science and Technology, vol. 47, no. 1, pp. 1–9. ACM (2010)
22. Marchionini, G, Geislerand, G., Brunk, B.: Agileviews: a human-centered framework for
interfaces to information spaces. In: Proceedings of the ASIS Annual Meeting, vol. 37 (2000)
23. Singh, J., Zerr, S., Siersdorfer, S.: Structure-aware visualization of text corpora. In: Proceed-
ings of the 2017 Conference on Conference Human Information Interaction and Retrieval
(CHIIR 2017), pp. 107–116. ACM, New York, NY, USA (2017). https://doi.org/10.1145/
3020165.3020182
24. Bandura, A.: Self-Efficacy. The Exercise of Control. W.H. Freeman and Company, New
York. Emory University, Division of Educational Studies, Information on Self-Efficacy: A
Community of Scholars
25. Maier, S.F., Seligman, M.E.: Learned helplessness: theory and evidence. J. Exper. Psychol.
Gen. 105(1), 3–46 (1976)
How Do Search Engines Shape Reality?
Preliminary Insights from a Learning
Experience
Abstract. More and more often, search engines are used by students as a tool
to access information on the Web. This has the potential to affect the learning
activities conducted by students. In this paper, the project “In WWW veritas?” is
presented. The project has the following objectives: (1) to investigate how searches
carried out on the Web through search engines can lead to different results based
on different criteria; (2) to increase students awareness on how search engines
“filters” can work and thus lead to a different perception of reality; (3) to stim-
ulate critical thinking in the use of searching tools on the Web to fully exploit
their potential. The project involved students of a high school in northern part of
Italy. They examined the results presented by a popular search engine on selected
controversial topics and tried to support or contrast the different points of view
through a role-playing game. A qualitative and quantitative analysis of results
showed an increased students’ awareness on the presence of filters through which
the search engines provide information. Moreover, the activities undertaken into
the project were also effective in developing critical thinking processes.
1 Introduction
Increasingly, search engines are used by students as a tool to acquire information, also
in support of learning. On the one hand, the use of search engines has facilitated the
retrieval of information on the net making it possible to access a multitude of contents.
On the other hand, search engines act as a filter between users and the Web, defining the
most relevant information which has to be proposed to the users.
The algorithm that determines the relevance of a given website in relation to a
specific search is based on several criteria. Although each search engine adopts a different
approach to determine the relevance of the results, and such approaches are generally
covered by trade secrecy, some of the parameters that significantly influence the results
are now known, and common to the most popular searching tools [1].
Furthermore, the same companies that develop search engines tend to make public
some of these parameters (albeit in an abstract form avoiding to specify the implementa-
tion details), highlighting the advantages for the users who can thus receive customized
answers; specific to their needs and habits.
The location from which a person connects, the language set in the browser, the type
of device used (desktop or smartphone), and navigation history are just some examples
of factors that can influence the behaviour of the search engines. Consequently, it is
quite common that people who query the same search engine with the same keywords
in different conditions, will get different results. As mentioned, these differences are
very often motivated by the introduction of improvements in usability namely, the user
experience. A user perceives useful to get as results on his smartphone only those sites that
are correctly displayed on the device; likewise, obtaining results for hotels or restaurants
that are nearby (therefore based on the user’s position) is a very useful feature for
travelers. Similarly, a user who connects from Italy and who has set the language of his
browser in Italian, will find useful to obtain pages mainly in Italian [2].
This ability of search engines to filter the contents of the Web has paved the way for
new research opportunities on social dynamics. Researchers from the Digital Methods
Institute of Amsterdam [3, 4] have shown how search engines provide different results
on particularly sensitive topics based on the country in which the search is conducted. In
[4] Rogers analysed the search results in different localized versions of search engines
in relation to the theme of “rights”. The results of this study showed that, in general,
different types of “rights” have different relevance in the various countries, certain types
are significant only in some specific countries, thus reflecting a different sensitivity to
the different type of “rights” in the world context. In this perspective, search engines
provide an effective analysis tool to support social research [5, 6].
One of the key features that influence the presentation of results is the number of
times a specific page, listed amongst the results, has been visited by the users. Also, the
previous search activities carried out by the user play a fundamental role. Indeed, users
can benefit from the behaviour of the algorithm implemented by the search engine. In
the first case, the user will have access to the most popular pages, those that are chosen
by the majority of users. In the second case, the user will be able to obtain personalized
search results, since the search engine will be able to infer what s/he already thinks about
a topic (according to her/his previous navigation history), and consequently it will filter
only the contents of interest of the user, thus contributing to strengthen the “bubble”
effect [7].
These mechanisms make search engines not only tools for retrieving information, but
environments in which the user interacts with content and information. In these terms,
it is simplistic to consider the outcomes of the searching process merely a list of results.
Students, who increasingly use the search engines as a first approach to the Web and
as a tool to forge their understanding on a new topic, are very often unaware of these
mechanisms, thus taking the risk to access polarized information on a specific point of
view, which does not support the development of critical thinking.
372 D. Taibi et al.
The present project “In WWW veritas?” aims to study how search engine “filters” can
literally “lead” students to misleading conclusions and theses; thus, inducing different
perceptions of reality.
Providing greater awareness in the use of search engines, they will enable students
to exploit the full potential of these tools, controlling them, avoiding to be controlled by
them. In parallel, from a didactic point of view, the project allows teachers to introduce
issues related to the reliability of Web sites and information on the Net, and “fake news”,
thus promoting the key competences necessary for media literacy [8].
“In WWW veritas?” is a project that combines both didactic and research aspects. The
project involved four classes of the last three-year of the high school located in a city in
the northern Italy for a period of about 3 months. The project involved 10 teachers and
about 100 students, for a total of 10 h of activity, 5 in classroom and 5 at home.
The project activities have been designed by researchers and teachers in order to
define:
• research methodology,
• technological tools to be used for the searching activities,
• tools for tracing students’ activities,
• questionnaire to collect students’ opinions.
In the initial phase, teachers presented the activities to be carried out, then they provided
technical instructions on the use of the tools, and finally they informed the students about
the data related to their online activities that would be collected. A privacy statement
indicating data management regulations has been delivered to students (Action 1.1).
Afterwards, the teachers discussed with students the topics under examination. Con-
troversial topics were selected in order to foster a debate amongst students. Examples
of keywords and topics on which the searches have been carried out were as follow:
selfie syndrome, holocaust, vaccines, homeopathy, N.G.O., horoscope, flat earth. Once
the topic to be explored online was chosen, the students carried out a search with their
computer at home (Action 1.2).
Phase 2 started with teachers discussing the individual results obtained by students during
their search at-home, highlighting:
The presence of different points of view was highlighted, and students were invited
to report and discuss some of the results they assumed to be of particular interest. The
teacher also guided students in the “critical reading” of the search engine results, by
providing examples based mainly on the results obtained by the students themselves. At
the end of this task, students were divided into two groups. Each group has to support a
point of view about the topic and to argue with the other group, sustaining an opposite
perspective on the same topic. It is important to note that students did not have to agree
374 D. Taibi et al.
necessarily with the point of view “sustained” by the group, but they have to play a
proper role-playing game (Action 2.1).
In the next action (Action 2.2), students used the devices (computers, tablet and
smartphones) available at the school laboratories to perform online searches by using
the same keywords used at home. At the end of the search sessions, students filled in
an ad-hoc questionnaire, designed to annotate different aspects of their search (e.g. how
many search results sustaining their point of view they obtained). The ultimate goal was
that of promoting a critical reading of the results. Subsequently, the teacher presented
various strategies to modify the keywords to be used in the search, such as: the use of
synonyms that have a relationship with the starting word (e.g.: vaccines and diseases), or
the use of keywords that describe concepts belonging to the same category (for example
in the case of vaccines, immunology). In Action 2.3, the teacher invited students to
carry out new searches by changing the keywords according to the strategies suggested,
with the aim of finding Web pages containing information in favour of the hypothesis
supported by the group the student belongs to. At the end of this search session, students
filled in another questionnaire to compare the results with the ones obtained in the
previous search sessions. Finally, in Action 2.4, students were asked to perform the
search using exactly the same keywords used at home and in the first search session at
the school laboratory. This task had the aim of highlighting the potential polarization of
the results. Indeed, the search activities performed by the two groups of students aimed
to sustain a specific point of view, led the search engine to produce polarized results. A
final questionnaire was filled in by the students in order to annotate any of this condition.
The third and final phase of the project consisted in data analysis. In this phase, the
teacher guided students to reflect (initially within the group) on the differences between
the first and second search sessions and the changes detected at the end of the third
search session. Finally, the students had to defend the point of view supported by the
group they belong to. At the end of this phase, the teacher stimulated reflection together
with the class on the reasons why different search results come up while using the same
search engine (in this case Google). Other search engine alternatives (e.g. Duck Duck)
were also presented to the students, in order to induce them to compare the differences
in the algorithms used.
3 Discussion
3.2 Results
The analysis of the results highlighted interesting data regarding the navigation style of
the students. During the experimentation 40% of students visited only one of the results
proposed by the search engine, and only 24% visited six or more results (Fig. 2).
Regarding the number of result pages viewed by students: 82% of students viewed
only the first page (up to 10 results as shown in Fig. 3). This means that the majority of
376 D. Taibi et al.
students did not go beyond the first 10 results proposed on the first page by the search
engine.
These preliminary results are particularly interesting from the educational point of
view, because they lead to a reflection on the limits of student interactions with search
engines, thus pointing out the needs for a more effective literacy on the use of search
engines.
4 Conclusions
The project “In WWW veritas?” has the aim of promoting a more aware use of search
engines for finding information on the Web. Specifically, teachers identified the following
key points that stimulated reflection within the classes:
• the “bubble” effect: the type of search engine response varies according to different
parameters (place from which the search is performed, language used for the search)
and in relation to the “history” of previous searches
• the reading style: which element of a singular page the students look at and in which
order
• the searching style: the keywords used, the number of pages viewed (less than ten or
more) and/or explored, and the time spent exploring a singular page
• the reliability of the source that is based on specific descriptors such as authorship,
frequency of the update, graphical quality
• the propagation of information on the net.
Furthermore, the search styles of students were analysed, highlighting how very
often they do not interact with the search engine results properly, thus increasing the risk
of accessing only that part of contents filtered by the search engine.
The interviews conducted at the end of the project activities with students and teach-
ers, resulted in a considerable interest. This result encourages to widen the number of
participants, extending the experimentation to schools of different order and types and
in different areas (in order to take into account also the localization of the results). The
How Do Search Engines Shape Reality? 377
Acknowledgments. Students and teachers of the high school “G. Peano - S. Pellico” in Cuneo
(Italy).
References
1. Langville, A.N., Meyer, C.D.: Google’s PageRank and Beyond: The Science of Search Engine
Rankings. Princeton University Press, Princeton (2011)
2. Zhang, J., Lin, S.: Multiple language supports in search engines. Online Inf. Rev. 31(4), 516–
532 (2007)
3. Rogers, R.: Digital Methods. MIT Press, Cambridge (2013)
4. Rogers, R., Jansen, F., Stevenson, M., Weltevrede, E.: Mapping democracy. In: Global Informa-
tion Society Watch 2009, pp. 47–57. Association for Progressive Communications and Hivos
(2009)
5. Fulantelli, G., Marenzi, I., Ijaz, A., Taibi, D.: SaR-Web – a tool to support search as learning
processes. In: Proceedings of the 2nd International Workshop Search as Learning (SAL) 2016
at the ACM SIGIR 2016 (2016)
6. Taibi, D., Fulantelli, G., Marenzi, I., Nejdl, W., Rogers, R., Ijaz, A.: SaR-WEB: a semantic
web tool to support search as learning practices and cross-language results on the web. In:
IEEE 17th International Conference on Advanced Learning Technologies (ICALT), Timisoara,
pp. 522–524 (2017)
7. Cardon, D.: Che cosa sognano gli algoritmi. Le nostre vite al tempo dei big data. Mondadori
Università. De Carolis, C. (Translator) (2016)
8. Averame, M.C.: Riconoscere le fake news in classe. Percorsi per una comunica-zione
consapevole in rete. Collana “insegnare nel XXI secolo”. Pearson (2018)
Author Index